risk management and the pension fund industry · such strategies are not only driven by funda- ......

As financial markets develop, a varietyof nonbank institutions, such asinsurers, pension funds, mutualfunds, and hedge funds, have been

increasing their exposure to market andcredit risks. This chapter is the second in aseries on the financial stability implications ofthis reallocation and transfer of risk, followingthe chapter, “Risk Transfer and the InsuranceIndustry,” in the April 2004 GFSR. This chap-ter focuses on pension funds, as significantinstitutional investors.

Pension funds have an impact on the stabil-ity of financial markets in several ways, mostsignificantly through their investment behav-ior. The global size and projected growth ofthe pension fund sector mean that thisinvestor class can move markets in its ownright. Any sizable reallocation of assets, saybetween fixed income and equities, couldhave a bearing on financial market stability.Such strategies are not only driven by funda-mental business models but also by cyclicalfactors and risk management considerations,as well as by official policies in areas such astaxation, regulation, and financial accounting.The changing needs of aging pension fundmembers also have a longer-run impact. Assuch, an analysis of the pension funds’ impacton financial stability will have to cover all ofthe above elements.

This chapter looks at the longer-term chal-lenges pension funds face as populations age,and the key issues to address in order toenhance their risk management practices andtheir role as long-term investors. The chapterfocuses primarily on Japan, the Netherlands,

Switzerland, the United Kingdom, and theUnited States, where funded pension plansare most developed. The size of pension sav-ings in these countries, their projected growth(whether managed by the state, corporations,or individuals), and the more recent develop-ment of funded pension schemes in othercountries, such as France, Germany, and Italy,highlight the fast-growing importance of pen-sion funds for international capital marketsand to financial stability.

How pension funds manage risk has a veryimportant bearing on the distribution offinancial and other risks among the differentsectors of the economy. As employers andgovernments have become more aware ofthe funding challenges pension funds facefrom aging populations, and more consciousof the investment risks involved in fundedpension plans, they have sought to managethat risk in a variety of ways. Reductions instate pension benefits in most countries, andmovements from defined benefit (DB) todefined contribution (DC) pension plans bymany businesses, have increasingly trans-ferred retirement risk (including investment,market, longevity risks, etc.) to the house-hold sector.1

National pension systems are typically repre-sented by a “multi-pillar” structure, with thesources of retirement income derived from amixture of government, employment, andindividual savings. A variety of definitions ofthe pillars are used in academic literature, gen-erally dependent on the purpose of each study.In this chapter, we identify and discuss threepillars, based primarily on the source of

81

CHAPTER III RISK MANAGEMENT AND THE PENSION FUND INDUSTRY

81

1Defined benefit schemes are those in which the employer commits to provide specific benefits related to anindividual’s wages and length of employment, while under defined contribution plans the commitment is to makespecific contributions to a pension fund, with the benefits dependent on the level of contributions to the schemeand the investment return. For definitions of other pension terminology see the glossary.

savings (i.e., government, employment, orindividual): Pillar 1—the state, often a combi-nation of a universal entitlement and an earn-ings-related component; Pillar 2—occupationalpension funds, increasingly funded, organizedat the workplace (e.g., DB and DC, and newerhybrid schemes); and Pillar 3—private savingsplans and products for individuals, often tax-advantaged. These are the definitions com-monly used by industry participants andanalysts, and are particularly suitable for ourfocus on risk transfer.2

This chapter primarily focuses on Pillar 2,as collective funds organized through theworkplace. Our focus reflects the role of Pillar2 funds as a major institutional investor class.The design of Pillar 1 programs will not bediscussed, as this is primarily a fiscal issue,although it should be noted that in someadvanced economies, such as Japan, France,and Canada (and certain developing econo-mies), some public sector schemes are (atleast partially) funded.3 This chapter will onlybriefly discuss Pillar 3 and efforts by some gov-ernments to encourage long-term retirementsavings generally, as we plan to discuss thefund management industry and householdsector in more detail in the March 2005 GFSR.Indeed, the economic characteristics of DCplans, including their allocation of risk, arevery similar to Pillar 3, and this chapter focusesmore on the management of DB plans andthe forces moving funds from DB to DC,rather than on the management of DC plansthemselves.

Pillar 2 funds can enhance financial stabilityby acting as a stable, long-term investor base;however, increasingly a variety of factors areinfluencing their structure, investment behav-ior, and management of risks. These factors,and how we arrived at a point many call “a

pensions crisis,” are discussed in the chapter.Similar to our previous work, we have high-lighted influencing factors, such as marketcharacteristics, regulatory and tax policies,and accounting principles. Finally, we look atdifferent investment strategies and risk man-agement approaches, and how these may helppension funds take a long-term perspective,and thereby support financial stabilityobjectives.

Why Pension Funds Are Important forFinancial Stability

An Aging Workforce

The importance of pension savings hasincreased dramatically in recent years, particu-larly as populations mature. Historically, lowproportions of pensioners in the overall popu-lation and the relatively larger workforce fromthe “baby boom” generation kept the burdenof pension outlays somewhat modest. DBschemes seemed a manageable and evenattractive (due to benefit deferral) proposi-tion to many companies. But as populationsage, the relative size of pension liabilities andinvestment risk grows. The growth in liabilitieshas been greater than expected, as increasesin longevity have consistently exceeded earlieractuarial forecasts. Questions of managingand maintaining funding levels have becomemore urgent, and some pension providers willfind it increasingly difficult to meet their pay-ment obligations according to their existingbenefit structures. For policymakers, the rela-tive burdens and merits of each of the threepillars are increasingly a prominent topic ofpolitical and social debate.

Advanced economies are confronted with avariety of retirement challenges associated


82

2Another definition used in pension studies, particularly for emerging markets, was first developed in WorldBank (1994). It describes Pillar 1 as “non-contributory state pension,” Pillar 2 as “mandatory contributory,” andPillar 3 as “voluntary contributory.” This definition has been most useful for considering questions of social safetynets, redistribution of income, and related issues.

3Many emerging market economies also wholly or partially fund public sector pension schemes. Emerging mar-ket pension issues will be discussed in future GFSRs.

with population aging, reflecting in part twolong-term trends:4

• Increasing longevity. In recent decades, lifeexpectancy at birth has consistentlyincreased in all advanced economies, froman average of about 68 years in the postwarperiod to 78 years today (Table 3.1 andFigure 3.1), and is projected to reach 80years or more by 2020. Importantly for pen-sion costs, life expectancy after age 65 is alsorising steadily, from 18 years currently, to aprojected 20 years or more in 2020 in theUnited States and some selected Europeancountries, and rising steeply in Japan(Figure 3.2).

• Low and declining fertility rates. In advancedeconomies between the early 1950s and thelate 1990s, fertility rates have dropped fromabout 2.8 to 1.7 children per woman, andare below the replacement rate in mostadvanced economies, except the UnitedStates.While population aging is a global phenom-

enon, it is happening rapidly in some coun-tries. The aging trend is particularly visiblein Italy, Japan, and Switzerland, where the

WHY PENSION FUNDS ARE IMPORTANT FOR FINANCIAL STABILITY

83

Table 3.1. Life Expectancy at Birth: Estimatesand Projections(In years)

1955 1980 2000 2020 2050

United States 68.9 73.3 76.2 78.7 81.6Japan 63.9 75.5 80.5 84.3 88.1Selected European

countries1 67.6 73.3 77.7 80.5 83.2

Sources: United Nations, World Population Prospects: The 2002Revision; and IMF staff estimates.

1Weighted average for France, Germany, Italy, the Netherlands,Switzerland, and the United Kingdom; weights are based on thecountries’ total population data for 2000.

60

65

70

75

80

85

90

United States

Japan

Selected European countries1

1955 65 75 85 95 2005 15 25 35 45

Figure 3.1. Life Expectancy at Birth(In years)

Sources: United Nations, World Population Prospects: The 2002 Revision; and IMF staff estimates.

1Weighted average for France, Germany, Italy, the Netherlands, Switzerland, and the United Kingdom; weights are based on the countries’ total population data for 2000.

15

17

19

21

23

25

27

United States

Japan

Selected European countries1

1990–95 2000–05 2010–15 2020–25 2030–35 2040–45

Figure 3.2. Remaining Life Expectancy at Age 65(In years)


1Weighted average life expectancy at age 65 for France, Germany, Italy, the Netherlands, Switzerland, and the United Kingdom; weights are based on the countries’ total population data for 2000.

4We have explicitly excluded the health and medicalissues from the scope of this study, in order to focus onfunded pensions. However, health and medical costsare rising rapidly in all the mature markets, and to theextent that such private schemes are funded at all, thefunding levels are significantly lower than pensions.(For example, for companies in the Standard & Poor’s500 at the end of 2003, the average funding levels wereapproximately 87 percent for pension liabilities and 15percent for medical and health care plans.)

median age is already above 40 years today,and projected to approach 50 years by 2020(Figure 3.3). Moreover, national differences inmedian age are projected to widen in thecoming years.

A direct implication is the continuedincrease in the dependency ratio—the ratio ofpensioners to working age population. Thedependency ratio is currently about 20 percentin Europe, Japan, and North America, and isprojected to increase rapidly once the “babyboom” cohort begins to reach retirement agearound 2010. By 2030, this ratio may reach 30percent in North America, 45 percent inEurope, and 55 percent (and rising rapidly) inJapan (Figure 3.4). The demand for retirementincome relative to contributions from workingincome will be proportionately greater, andthis pressure will be felt by private companies(particularly in older or declining industries)as well as by public/state programs.

Policymakers have started to address thesechallenges and to rethink their pension sys-tems. Thus far, pension reforms frequentlyhave been aimed at reducing the generosity ofexisting systems in various ways: reducing ben-efits, increasing contributions (e.g., taxes topay for state pensions), redefining risk sharingbetween sponsors and beneficiaries, andincreasing the retirement age. Given the scaleof the problem, it is likely that actions on sev-eral of these fronts will be needed.5 Increasedfunding of pension obligations, by both thepublic and private stctors, and greater retire-ment savings by individuals (Pillar 3), areincreasingly part of the solution.

Pension Funds Are Significant Investors in GlobalFinancial Markets

Funded pension plans’ size and importanceto financial markets vary sharply between dif-ferent countries. The countries we have stud-ied can be broadly classified into two groups:


84

20

25

30

35

40

45

50

55202020001970

Netherl

ands

Franc

e

United

Kingdo

m

German

y

Switzerl

and

Italy

Japa

n

Canad

a

United

States

Figure 3.3. Median Age of Population, by Country1

(Medium fertility variant)

Source: United Nations, World Population Prospects: The 2002 Revision.1Countries are shown in increasing order of the median age of population in 2000. The

United Nations’ medium fertility variant assumes that fertility levels converge to 1.85 births per woman in all countries.

0

10

20

30

40

50

60

70

80

Japan

United StatesSelected European countries2

Figure 3.4. Dependency Ratio for Selected Countries1

(In percent)


1Population aged 65 and over as a percentage of population aged 15 to 64.2Weighted average dependency ratio for France, Germany, Italy, the Netherlands,

Switzerland, and the United Kingdom; weights are based on the countries’ total population data for 2000.

1950 60 70 80 90 2000 10 20 30 40 50

5See, for example, Turner (2003) and Moody’sInvestors Service (2004).

those where pension assets represent morethan 60 percent of GDP, including theNetherlands, Switzerland, the UnitedKingdom, and the United States; and thosewhere pension assets represent less than 20percent of GDP, including France, Germany,Italy, and Japan (Table 3.2).6

In a number of countries, pension fundsare the largest class of institutional investor.Pension funds represent about 50 percent ormore of institutionally held assets in theNetherlands and Switzerland; over 33 percentin the United Kingdom and the UnitedStates; and about 20 percent in Japan. Theproportion remains negligible in countrieswhere private pension savings are not welldeveloped or are chiefly managed by insur-ance companies, for instance in France andGermany (Table 3.3).

The investment behavior of pension fundscan have a significant effect on markets, asthey hold a large and growing proportion ofoverall financial assets. As of the end of 2001,pension funds in the United Kingdom andthe United States held domestic equities equalto 18 and 22 percent, respectively, of totaldomestic equity market capitalization (Table3.4). Meanwhile they held domestic bonds(both credit and government securities)equivalent to 11 and 9 percent, respectively, oftotal domestic bond market capitalization. Inthe Netherlands, pension funds’ total equityallocation (both domestic and foreign) equals36 percent of the country’s domestic equitymarket capitalization, and Swiss pensionfunds’ total bond allocation (domestic andforeign) equals 59 percent of the domesticbond market capitalization, leading pension

funds from both countries to invest substantialproportions abroad. In contrast, pensionfunds’ relative holdings in Germany, Italy, andJapan are much smaller.7 But with these andother countries moving toward increasedfunding of pension liabilities, the global pen-sion fund industry and its impact on financialmarkets can be expected to grow.

Changes in Pension Funds’ Asset AllocationsCould Impact Financial Markets

There is an ongoing debate on the meritsof pension funds holding bonds versus equi-ties, raising the question of whether bond andequity markets could be impacted by majorportfolio reallocations. Equity allocations arecurrently as high as 50–70 percent in manypension funds in Japan, the Netherlands, theUnited Kingdom, and the United States(Figure 3.5). Fund managers, pension consult-ants, and market analysts increasingly believethat regulatory and accounting changes(under consideration or recently adopted)could trigger a significant reallocation of pen-sion assets from equities into bonds, as spon-sor companies seek to reduce funding riskand accounting volatility (see the section,“Asset Allocation and Risk Management,” laterin this chapter). An immediate or short-termreallocation from equities could have a signifi-cant impact on financial markets and assetprices in the short term. However, such a shiftwould seem unlikely given the reluctance ofmany pension fund managers to move fromequities to bonds (or pursue more closelymatched risk management strategies) whilethey remain significantly underfunded.8 The


85

6The Japanese figures exclude assets held by the Pillar 1 Government Pension Investment Fund (GPIF).Although this public pension scheme is a pay-as-you-go system, it has accumulated a surplus from contributionsworth ¥150 trillion, invested in government bonds, equities, and foreign securities.

7In Japan, in addition to occupational pension funds, the GPIF’s equity holdings amount to 3 percent of domes-tic stock market capitalization.

8In the appendices to CIEBA (2004), a Morgan Stanley research report estimates that an abrupt reallocationcould lead to a temporary 10 to 15 percent reduction in U.S. equity prices and a 75–150 basis point flattening ofthe U.S. government bond yield curve, while a Goldman Sachs paper estimates only a 1 percent reduction in equityprices and a 10 basis point reduction in long-term yields.


86

Table 3.2. Asset Allocation of Autonomous Pension Funds1

(In percent of total financial assets of pension funds, unless otherwise noted)

1992 1995 1998 2001

GermanyCash and deposits 1.5 1.8 1.8 2.0Bonds 49.5 54.9 55.8 57.4Equities 0.2 0.0 0.0 0.1Loans 48.1 43.0 42.2 40.5Other 0.8 0.2 0.2 0.1

Memorandum items:Financial assets (in billions of U.S. dollars) 56.6 65.3 69.3 60.5Financial assets (in percent of GDP) 2.9 2.7 3.1 3.3

ItalyCash and deposits 32.4 38.5 45.4 36.0Bonds 42.2 33.3 36.1 40.5Equities 0.1 2.2 0.9 6.8Loans 0.0 0.0 0.0 0.0Other 25.3 26.1 17.7 16.7


Japan2

Cash and deposits and other 2.2 1.9 2.5 2.7Bonds 28.9 27.1 30.7 31.5Equities 19.4 25.3 46.9 52.3Loans 8.9 5.5 2.2 1.5Insurance 40.3 39.9 17.7 12.1

Memorandum items:Financial assets (in billions of U.S. dollars) 416 634 619 611Financial assets (in percent of GDP) 10.7 13.1 13.9 16.0

NetherlandsCash and deposits 1.9 2.1 1.5 1.5Bonds 22.8 27.4 33.5 34.7Equities 17.8 27.2 40.1 49.5Loans 48.3 35.7 19.1 8.8Other 9.2 7.6 5.8 5.4


Switzerland3

Cash and deposits 10.0 11.3 10.7 8.5Bonds 40.5 36.9 35.5 35.9Equities 13.1 25.5 31.9 39.0Loans 34.8 23.4 19.3 13.8Other 1.6 2.9 2.6 2.9


United KingdomCash and deposits 3.6 4.0 4.4 3.3Bonds 9.9 13.4 15.8 14.5Equities 74.8 70.8 66.8 63.5Loans 0.1 0.0 0.0 0.0Other 11.6 11.7 13.0 18.8

Memorandum items:Financial assets (in billions of U.S. dollars) 552.4 759.7 1,136.5 954.0Financial assets (in percent of GDP) 52.7 68.2 79.3 66.4

impact of a more gradual reallocation is moredifficult to assess, especially as broaderchanges in the risk management practices ofpension funds can be expected in the comingyears.

Pension fund demand could have a particu-larly pronounced impact on certain assetclasses. Pension funds are increasingly focus-ing on asset-liability management (ALM) (i.e.,ensuring that liabilities are sufficiently cov-ered by suitable assets) and in particular therelative duration of assets and liabilities. Manymarket participants highlight the relativelyshort supply for this purpose of long-termbonds (i.e., 20 to 30 years or longer), and par-ticularly inflation-indexed bonds (see Table

3.5). At present, even a relatively modest real-location of pension assets into these long-termsecurities would overwhelm the market, asliquidity constraints could lead to significantshort-term price volatility. Over time, however,the supply of long-term and inflation-indexedbonds may increase, possibly with governmentleadership, and we would expect pensionfunds to be a significant investor.

The potential for greater internationaldiversification by pension funds could alsohave a strong impact on international capitalflows. In particular, as populations age in themature markets and their need for retirementsavings grows, this creates potential demandto make additional investments in countries


87

United StatesCash and deposits 4.5 3.7 3.7 3.7Bonds 31.1 26.9 21.1 23.1Equities 46.5 54.3 62.5 59.8Loans 2.8 1.8 1.6 1.8Other 15.0 13.2 11.0 11.5

Memorandum items:Financial assets (in billions of U.S. dollars) 3,011.6 4,226.7 6,231.9 6,351.3Financial assets (in percent of GDP) 50.0 57.1 71.0 63.0

Sources: OECD Institutional Investors Yearbook; Japanese Pension Fund Association; and Bank of Japan, Flow of Funds.1Occupational and personal pension funds, legally separated from the plan/fund sponsor taking the form of either a special purpose legal

entity (a pension entity) or a separate account managed by financial institutions on behalf of the plan/fund members.2Asset allocation shares are those of Employee Pension Funds only. Memorandum items include all pension fund assets.3For 1995 and 2001, data refer to 1996 and 2000, respectively.

Table 3.2 (concluded)

1992 1995 1998 2001

Table 3.3. Financial Assets of Institutional Investors, 20011

(In percent of total financial assets, unless noted otherwise)

Total Financial Investment InsuranceAssets2 Companies3 Pension Funds Companies Other

France 131.8 47.7 . . . 52.3 0.0Germany 81.0 44.9 4.1 51 0.0Italy 94.0 35.6 4.7 23.7 35.9Japan 94.7 10.0 19.5 63.7 6.8Netherlands4 190.9 11.9 55.0 32.3 0.8Switzerland5 232.7 13.8 48.8 37.4 0.0United Kingdom 190.9 14.4 34.8 50.8 0.0United States 191.0 34.3 33.0 21.2 11.5

Source: OECD Institutional Investors Yearbook.1Institutional investors are insurance companies, investment companies, and pension funds.2In percent of GDP.3Open-end and closed-end investment companies.4For 2001, excluding nonlife insurance.5For 2001, including total assets of pension funds.

with younger labor forces (in particular,emerging markets), and raises questionsabout the ability of those markets to absorbsubstantially greater flows.

The Funding ChallengeThe debate over the design and asset alloca-

tion of pension funds has taken on moreurgency as the industry has swung from over-funded to underfunded status in recent years.These factors have focused attention on theinvestment and other risks associated with tra-ditional DB plans. This has led to a closer con-sideration of the merits of different assetclasses in matching pension liabilities andaccelerated the industry’s consideration of DCand hybrid pension plan alternatives to tradi-tional DB schemes.

How Pension Funds Became Underfunded

Several factors have led pension funds tobecome underfunded in recent years.9 Thissection focuses on the rising level of DB planpromises, especially relative to contributions,and on the impact of falling equity marketsand interest rates.


88

0

20

40

60

80

100

120

140OtherCash and deposits

LoansBondsEquities

German

yIta

ly

Japa

n2

Netherl

ands

United

Kingdo

m

United

States

Switzerl

and3

Figure 3.5. Asset Allocation of Autonomous Pension Funds, 20011

(In percent of financial assets of pension funds)

Sources: OECD Institutional Investors Yearbook; Japanese Pension Fund Association; and Bank of Japan, Flow of Funds.

1Occupational and personal pension funds, legally separated from the plan/fund sponsor taking the form of either a special purpose legal entity (a pension entity) or a separate account managed by financial institutions on behalf of the plan/fund members.

2For Japan, “other” refers to insurance sector; and “cash and deposits” refer to cash, deposits, and other. Allocations are those of Employee Pension Funds (EPFs) only.

3For Switzerland, data refer to 2000.

Table 3.4. Pension Fund Holdings Compared with theSize of Domestic Market, 2001(In percent)

Equities1 Bonds2_____________________ _____________________Domestic International Domestic International

Japan 7.4 . . . 3.2 . . .Netherlands 6.5 29.4 15.2 23.3Switzerland3 6.9 5.8 38.1 21.1United Kingdom 18.1 9.8 11.2 3.4United States 22.4 5.1 8.7 0.2

Sources: OECD Institutional Investors Yearbook; BIS; Bank of Japan,Flow of Funds; World Federation of Exchanges; Datastream; UBS GlobalAsset Management; and IMF staff estimates.

1Holdings of equities as a percentage of total domestic marketcapitalization.

2Holdings of securities over one year in maturity as a percentage of totalpublic and private domestic debt securities outstanding.

3Data refer to 2000.

9See, for instance, IMF (2003).

Pension funds in North America and partsof Europe historically have held significantamounts of equities in their portfolios. Thisreflected a belief in the greater long-runreturns expected from equities compared withbonds. In the United States, this also partlyreflected the interpretation of the “prudentperson” rule introduced as part of theEmployee Retirement Income Security Act(ERISA) in 1974, which in part requires“diversifying investments . . . so as to minimizethe risk of large losses, unless under the cir-cumstances it is clearly not prudent to do so,”which led many pension funds to more sys-tematically diversify across asset classes.10

During the 1990s, as equity prices rose, thefunding ratio of many DB plans rose wellabove 100 percent. While accounting and actu-arial smoothing of market valuations reducedthe immediate impact of equity prices onfunding ratios, the steady rise in equity pricesfed through over time (Figures 3.6 and 3.7).In some cases, the “overfunding” was furtherexaggerated by the use of above-market or rel-atively fixed discount rates for funding ratiocalculations, even as market rates for bondsfell throughout most of the 1990s.11

Moreover, projections of future returns,based largely on recent performance, furtherboosted calculated funding ratios by extrapo-lating forward these current strong equitymarket returns.

Sponsor companies often acted to “realize”these gains, thereby weakening the capacity ofpension funds to absorb future shocks. Inparticular:• Many sponsor companies reacted to their

pension fund’s overfunding (both real andexaggerated) by reducing or eliminatingcontributions. Sponsor companies were ableto reduce their annual contributions (or, insome cases, tax regulation penalized furthercontributions) and in many cases take “con-tribution holidays” of a decade or more. Inother cases (such as in Switzerland or in theU.S. public sector), contributions byemployees were reduced as well.

• Companies with surplus pension fundingalso frequently increased the size and scopeof benefits, including through indexation.The costs of these benefit increases did notdirectly affect companies’ reported profits.But in practice they introduced permanentincreases in liabilities and greater risk to the

THE FUNDING CHALLENGE

89

Table 3.5. Selected Countries: Total Outstanding Long-Term Bonds(In billions of U.S. dollars)

United States United Kingdom France Italy Japan____________ ______________ ____________ ____________ ____________2000 2003 2000 2003 2000 2003 2000 2003 2000 2003

Corporate and government long-term bonds1 1,143 1,257 144 202 74 128 81 223 250 368Inflation-indexed government bonds2 115 166 99 139 12 59 1 11 . . . . . .

Memorandum item:Total pension fund assets (at end-2001) 6,351 954 . . . 47 711

Sources: U.S. Department of the Treasury; U.K. Debt Management Office; Agence France Trésor; Italy, Ministry of Economics and Finance;Japan, Ministry of Finance; Merrill Lynch; and OECD, Institutional Investors Yearbook 2003.

1Total amount of 10-year and above maturities. For the United Kingdom, France, and Italy, government bonds only.2For France and Italy, also includes bonds indexed on euro area inflation.

10Asset allocations became more similar in the U.S. pension fund industry after the adoption of ERISA, as the“prudent person” rule contributed significantly to a convergence in asset allocation between different pensionfunds. In 1970, the equity allocation of state and local government pension funds was 23 percent, whereas that ofprivate trusteed pension funds was 54 percent. By 2000, the allocations were much more similar, at 58 percent and48 percent, respectively.

11In Japan, the Netherlands, and Switzerland, fixed discount rates for liabilities were used, while U.K. discountrates allowed a large element of actuarial discretion and were typically set well above market rates (although a mar-ket-related element was increasingly used from the mid-1990s onwards).

financial strength of pension funds, withthe costs and risks further magnified byincreases in longevity beyond earlier actuar-ial projections. In some cases, generousearly retirement packages were used toincrease turnover in the workforce and tophase out DB plans and introduce new DC-style plans for younger employees.Japan also experienced overfunding, but

with a different timing. Overfunding devel-oped in the late 1980s during the asset marketbubble. However, poor returns in the 1990son both equity and fixed-income markets,together with returns of 5.5 percent requiredon Pillar 1 pension contributions managed byemployers, led to 66 percent of private pen-sion funds becoming underfunded by 1996.12

Overfunding briefly occurred again followingthe abolition of investment limits in 1996,which allowed an increase in equity holdings.(Under previous limits, pension funds wererequired to invest more than 50 percent oftheir assets in bonds, and less than 30 per-cent each in equity and foreign securities.)Japanese pension funds raised their allocationin equities to above 50 percent by 2000, andat that time 82 percent of Japanese funds wereoverfunded (Figure 3.8).13

Between 2000 and 2002, pension fundsworldwide became significantly underfunded.The equity market fall of 2000–02 sharply cutthe funding ratios of pension funds that, inmany cases, held equity allocations of 50 per-cent or more (see Table 3.2). Moreover, mar-ket interest rates, which increasingly werebeing used in some jurisdictions (such asJapan and the United Kingdom) as the basis


90

60

70

80

90

100

110

120

130

0

50

100

150

200

250

300

350

400

450

500

3

4

5

6

7

8

9

Unadjusted PBO funded ratio

10-year government bond yield(right scale; in percent)

S&P 500(left scale; January 1991 = 100)

1991

1991 93 95 97 99 2001 03

93 95 97 99 2001

Figure 3.6. United States: Ratio of Assets to Projected Benefit Obligations (PBOs) for the Fortune 500(In percent)

Sources: Hewitt Associates; and Bloomberg L.P.12In Japan, employers providing Pillar 2 pensions

in the form of Employee Pension Funds (EPFs) arealso required to administer (as agents) the govern-ment’s Employee Pension Insurance (EPI) for theiremployees, withholding contributions from employ-ees’ salaries and managing the funds to provide afixed return. In return, the EPFs are allowed to beoverfunded, with the profit or loss from investing EPIreturns absorbed into the overall EPF fundingposition.

13See Watson Wyatt (2003).

for discounting liabilities, fell significantly,thereby increasing the present value of liabili-ties and creating the “perfect storm” for pen-sion funds.14 In the United Kingdom, the shiftfrom contribution holidays to large annualcontributions was made all the more extremeby the fact that Minimum Funding Require-ment (MFR) thresholds began to dictate fund-ing policy at many firms. (The MFR fundingcalculation uses more market-related discountrates—i.e., at that time, lower rates—andhence larger valuations of liabilities, than pre-viously controlling actuarial funding calcula-tions.) Even the assets held in the form offixed-rate bonds failed to grow in value as fastas liabilities, largely because the average dura-tion of such assets was typically much shorterthan the duration of liabilities. By the end of2002, over 90 percent of pension funds inJapan, the United Kingdom, and the UnitedStates were underfunded, and the rise ininterest rates and equity prices since has ledto only a partial recovery (Figure 3.9).

The impact of falling equity markets andbond yields on asset and liability valuationswas significant. Figure 3.10 shows one estimateof the effect of valuation changes on pensionfunds in different countries and regions. Itillustrates the impact of market changes on ahypothetical pension funding ratio, assumingthat the fund started with a funding ratio of100 percent at the beginning of 2000, andthat it had a typical asset allocation and liabil-ity structure for that country or region.

Although the fall in equity values has beenmost often credited as causing the under-funded position of many pension funds, thefall in bond yields (and the greater use ofmarket-related discount rates for liabilities)has been at least as important. Given the typi-cally long duration of pension fund liabilities,changes in yields (and thus discount rates)have a major impact on the calculated valueof liabilities. In the United States, for


91

80

90

100

110

120

130

140

0

1

2

3

4

5

6

7

8

9

10

0

50

100

150

200

250

300

350

400

450

500

Funding ratio of pension funds

10-year government bond yield(right scale; in percent)

FTSE Eurotop 300(left scale; January 1991 = 100)

1991 93 95 97 99 2001

Figure 3.7. Netherlands: Funding Ratio of Pension Funds(In percent)

Sources: Netherlands Pension and Insurance Supervisory Authority; Van Ewijk, and van de Ven (2003); and Bloomberg L.P.

1991 93 95 97 99 2001 03

14See Hewitt Investment Group (2001) and Custis(2001).

instance, it has been said as a rule of thumbthat each 10 basis point change in the dis-count rate leads to a 1 percent change in pro-jected benefit obligations (PBOs) (Standard& Poor’s, 2004a). Meanwhile, a recent actuar-ial estimate suggested that the aggregateunderfunding of the 200 largest U.K. DBschemes would be eliminated by either a 30percent rise in equity prices or a 1 percentagepoint rise in bond yields (Aon Consulting,2004). This demonstrates the significant, andoften underappreciated, influence market-related discount rates can have on fundingratios.

Companies have only limited scope in theshort term to address their underfunding byreducing benefits or increasing contributions.Companies have had little room to reducerecently increased benefits—in fact, they weresometimes legally constrained from scalingback benefits (for instance, in the UnitedKingdom indexation up to a cap of 5 percentbecame a regulatory requirement in 1997).Weaker corporate profitability in 2001–02,and the ongoing decline in the financialstrength of older industries, also restricted theability of some sponsor companies to raisecontributions.

The deterioration in funding levels, and thequestions raised in some cases about the cor-porate sponsor’s ability to meet future obliga-tions, brought urgency to the debate aboutpension fund structures and strategies. Theviability of DB schemes has been questioned,as well as the appropriate risk sharing betweenemployer and employee (Pillars 2 and 3),public and private sector responsibilities, andrelated social and tax policy issues. The rapiddeterioration of funding ratios accelerated theshift to DC schemes, and led to the develop-ment of new approaches to pension andretirement programs.

The Move from DB to DC and Hybrid Plans

Even before the deterioration in marketconditions and funding levels, there was a


92

–2000

–1500

–1000

–500

0

500

1000

1500

2000

Overfunded

Underfunded

1991 93 95 97 99 2001

Figure 3.8. Japan: Employee Pension Funds1

(Number of funds)

Source: Pension Fund Association.1Fiscal years; before 1996, data shown on book value basis; for 1996 and after, data

shown on fair market basis.

0

10

20

30

40

50

60

70

80

90

20-30

%

40-50

%

60-70

%

80-90

%

100-1

10%

120-1

30%

140-1

50%

>150

%

Figure 3.9. United States: Distribution of CorporateDefined Benefit Pension Plans by Funding Ratio, 20031

(Number of pension plans)

Source: Wilshire, 2004 Corporate Funding Survey on Pensions.1Survey of 331 companies in S&P 500 Index.

2 0

8

19

43

69

79

49

36

15

61 2 2

growing belief that many DB schemes, as tra-ditionally constructed, may need to beredesigned. DB schemes had become lessflexible, in large part through greater bene-fits and increasing longevity. In addition, theDB structure may be less suitable as employ-ees become more mobile—in fact, newerindustries (often less unionized) and theirgenerally younger workforces favor DC pen-sion schemes, as more mobile employeesare attracted by the portability of pensionbenefits. The move to more market-basedaccounting principles has also increased theperceived volatility of DB plan balancesheets.

In the United States, the use of DC planshas been growing for 30 years. The introduc-tion of ERISA in 1974, the creation of thePension Benefit Guaranty Corporation(PBGC), which imposed insurance premiumson DB funds, the strengthening of fundingrequirements, and, for some firms, a desire toreduce contribution levels supported thegrowth of DC plans (Figure 3.11). Over time,many DB plans have closed to new employeesand/or frozen benefits at existing accrued lev-els, and shifted all employees to new plans. By1985, over 35 percent of assets under manage-ment (AUM) by U.S. private pension fundswere in DC plans. Since then, DC plans (e.g.,401(k) plans) have continued to increase inpopularity in the United States, reaching closeto 55 percent of AUM by 2000, and growingfurther since then.

In the United Kingdom, the trend wasinitially slower, but the recent introductionof FRS 17 and a fair value accounting frame-work (to be fully implemented by January2005) has accelerated the move away fromDB schemes. In 2000, 80 percent of activeparticipants in private sector pension fundsstill belonged to DB plans, but more recentinformation suggests that 60 percent ofDB schemes (weighted by the number ofemployees) are now closed to new members(Jackson, Perraudin, and Trivedi, forthcom-ing). Many U.K. firms are also taking the


93

40

50

60

70

80

90

100

110

120

United States

United Kingdom

Japan

Euro area

2000 01 02 03 04

Figure 3.10. Estimated Valuation Effects on Projected Benefit Obligation (PBO) Funded Ratios1

(In percent)

Source: Towers Perrin.1The series show the effect of asset and liability valuation changes on a hypothetical

pension fund in each country or region with a typical asset allocation and liability structure, and with a funding ratio of 100 percent at the beginning of 2000, without allowing for contributions to the fund during the period.

Defined contributionDefined benefit

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

1985 87 89 91 93 95 97 99 2001 03

Figure 3.11. United States: Assets Under Management of Private Pension Schemes(In billions of U.S. dollars)

Source: Board of Governors of the Federal Reserve System, Flow of Funds.

opportunity to cut contribution levels as theymove to DC.

In Japan, DC schemes have not grown veryquickly, despite reforms in 2001 to allow DCplans to complement or replace traditionalDB pension funds. Tax Qualified PensionPlans, as they exist today, will be progressivelyphased out by 2012, and replaced by newexternally managed DB and/or DC schemes.15

Companies with EPFs (see footnote 12) aregiven an option whether to transform them-selves into the new DB or DC schemes. Theincreasing mobility of employees, and theintroduction of a more transparent account-ing framework in 2000, which revealed fund-ing gaps on sponsor companies’ balancesheets, prompted these reforms. The recentscaling back of Pillar 1 (cutting the benefitlevel by, on average, 10 percent of final salary)has also encouraged the development of DCplans.16 However, the growth of DC plans alsohas been impaired by the limitation of taxdeductions for employers (¥432,000 per year,per employee), and the fact that employeesare not allowed to contribute to the new cor-porate DC schemes.

To date, the move from DB to DC schemeshas not altered asset allocations a great deal.In the United States, individuals participatingin DC plans have tended to allocate themajority of their funds to equity investments,which is not substantially different from DBplans (60 percent of AUM in DC plans havebeen invested in equity on average since 1990,compared with 53 percent for DB plans). Butthe shift to DC is important to financial mar-kets, including, among other reasons, becauseof its transfer of risk from sponsor companiesto households. It remains to be seen whether

the current asset allocation pattern will con-tinue, particularly as aging populationsapproach retirement age.

Despite these statistics, many consultantsargue that households often remain too riskaverse. Financial consultants advise individualsto hold relatively large allocations of higherrisk instruments, such as equities, in pensionsavings when they are young, and to graduallyswitch to assets with more stable values, suchas bonds, as they approach retirement.Consultants in a number of countries repeat-edly stated that, while it may be tax efficient tohold bonds within pension savings, house-holds generally hold too little investment riskoverall. This is especially so for youngersavers, particularly when looked at in the con-text of their overall savings (including othernon-pension savings). However, in recentyears a variety of “life cycle” savings productshave been developed, which address asset allo-cation and adjustment issues related toaging.17 (A broader discussion of householdsector savings will be discussed in the March2005 GFSR.)

The reconsideration of DB plans has alsoled to the development of “hybrid” pensionplans (see Box 3.1). Many sponsor companieshave sought to share market and longevityrisk, and to adjust benefits depending on busi-ness conditions, while still guaranteeing aminimum benefit to employees. Such hybridplans incorporate elements of both DB (as thesponsor makes contributions and bears atleast some investment or guaranteed returnrisk) and DC plans (as benefits are oftenexpressed in terms of an account balance andoften result in a lump sum payment atretirement).


94

15Tax Qualified Pension Plans are the second largest form of pension plan in Japan, after EPFs, and are so calledbecause they meet Corporate Tax Law conditions for tax exemptions on contributions.

16See IMF (2004b).17In the United States, “life cycle mutual funds” were first developed in the 1990s. In order to match the pre-

sumed and recommended changing risk tolerance of individuals during their life, such funds provide greaterrisk taking in the early years, before automatically and gradually adjusting the asset allocation to a more conser-vative approach (e.g., reducing equity and increasing fixed-income investments) as the individual approachesretirement.

The use of hybrid schemes is growing in theUnited States, Europe, and Japan. In theUnited States in 2000, 21 percent of PBGC-covered plan members belonged to hybridplans. We anticipate further growth of suchplans, but legal uncertainties and technicaldifficulties linked to conversion from tradi-tional DB schemes in some cases may slowtheir development in the near term. ManyEuropean companies are also developinghybrid Pillar 2 schemes, which give employerssome flexibility over the provision of inflationprotection and longevity risks (see below). InJapan, due to the greater inflexibility in DCplans (as legislated), many companies haveadopted “cash balance” plans as part of theiramendment of DB schemes following the2001 reforms.

Dutch regulatory proposals have also movedtheir system closer to a hybrid model, and theUnited Kingdom has reduced the degree ofrequired indexation. The planned regulatoryreforms (including the development of arisk-based capital system, described below)encourage the traditionally indexed DB Dutchpension system to make pension indexationexplicitly conditional on market conditions.Meanwhile, the United Kingdom has decided

to halve the cap on required inflation indexa-tion to 2!/2 percent.

New National Approaches to Pension Schemes

European countries that have been develop-ing Pillar 2 and 3 systems in recent years bene-fited from the experience of countries withmore established funded pension schemes.They have been conscious of the financialconstraints arising from an aging population,and new designs have generally followed a DCor hybrid plan approach.

Currently, the relative importance and con-tribution of Pillars 1, 2, and 3 differ signifi-cantly from country to country (see Table3.6). In countries such as the Netherlands,Switzerland, the United Kingdom, and theUnited States, the public pension systemoperates in part as a safety net, designed toprovide a basic pension income, while Pillars2 and 3 provide a much more significant con-tribution to retirement or replacementincome than in other countries.18 In contrast,in most continental European countries thestate has traditionally been the main sourceof retirement benefits (generally pay-as-you-go, or PAYG), and Pillars 2 and 3 are typically


95

Table 3.6. Sources of Retirement or Replacement Income(In percent of total income)

United UnitedGermany France Italy Netherlands Switzerland Kingdom States Japan

Public sources1 85 79 74 50 42 61 41 34Private/all other sources2 15 21 26 50 58 39 59 66

Memorandum item:Overall replacement rate (percent)3 82 79 80 78 81 69 67 75

Sources: Adapted from Börsch-Supan (2004); Employee Benefit Research Institute; Pensions Policy Institute; Japanese Ministry of PublicManagement, Home Affairs, Posts and Telecommunications, Survey of Household Economy; and IMF staff estimates.

1Pillar 1 includes France’s AGIRC/ARRCO, the U.K.’s State Second Pension Scheme (S2P), and Japan’s EPI.2All private sources of retirement income, including occupational pension income as well as income from financial assets (including income

from the reinvestment of lump sums paid by Pillar 2 schemes), use of bank deposits (particularly important in Japan), and earnings from work(in the United States, earnings from work are estimated to represent close to 20 percent of retirement income).

3Pension income, just after retirement, as a percentage of total income just before retirement, for an average two-person household; excludessources of income other than pensions.

18See, for example, Queisser and Vittas (2000).

underdeveloped. In Germany, for instance,while many employers have for years providedPillar 2 (traditionally DB) schemes, the bene-fit represents a modest share of aggregatepension income.19 In France, the earnings-related mandatory AGIRC/ARRCO system,20

even if managed and funded by contributionsfrom both employees and employers, is inessence an additional layer of Pillar 1 (some-

times referred as “Pillar 1A”). In Italy, theTrattamento di Fine Rapporto system, underwhich employers pay a lump sum when anemployee leaves the company, has long beenthe closest proxy to a Pillar 2 scheme, but hasrepresented only a small part of retirementincome. While the framework for new DCschemes was established in 1993 in Italy, DCplans have gained momentum only since


96

Hybrid pension plans, in essence, have somefeatures of defined benefit (DB) plans, butoften with a greater sharing of risks by benefici-aries. Similar to traditional DB plans, theemployer/trustee invests the plan assets andtypically bears some of the investment risk.However, hybrid plans also operate in many wayslike defined contribution (DC) plans, in thatthe employee typically has an individual accountand can receive the account balance either inannuity form or as a lump sum at separation,thereby assuming more longevity risk. Theportability and relatively earlier accrual ofbenefits typically provided by hybrid plans areoften very attractive to today’s more mobileworkforce. At the same time, hybrid plans pro-vide to employees some of the advantages ofDB plans in terms of guarantees and assurance.Indeed, the terminology is not always well-defined and some “hybrid” schemes in effectprovide defined benefits.

Hybrid plans take a variety of forms acrosscountries, for example:• In Japan, the United Kingdom, and the

United States, “cash balance plans” (CBPs)are the most common form of hybrid pensionplan. CBPs in those countries are plans in

which a fraction of an employee’s salary isdeposited in a notional account (or “cash bal-ance”). Notional accounts are used for record-keeping purposes only, as the funds are notinvested for each individual separately, but forthe plan as a whole. The benefits are usuallybased on an average rather than final salary,and may or may not contain a variable ele-ment related to market returns (in Japanthey reflect asset returns with minimum guar-antees), and accrue more evenly over anemployee’s career than under traditional DBschemes. In the United States, CBPs arelegally classified as DB plans, and as such areinsured by the PBGC.

• In Germany, the growth of hybrid plansreflects (in part) the impact of regulationson capital guarantees imposed on pensionfunds. New vehicles introduced under theRiester reform, including the Pensionsfonds,are required to guarantee a minimum benefitequivalent to principal protection. Similarly,since 2002, other vehicles, including thePensionskasse and Direktversicherung, alsoneed to provide such guarantees in order tobenefit from state subsidies and taxdeductions.

Box 3.1. Hybrid Pension Plans

19Pillar 2 pension benefits are estimated to represent approximately 5 percent of retirees’ overall income. At end-March 2003, 43 percent of private sector employees (46 percent in western Germany and 27 percent in easternGermany) were members of occupational pension schemes. Pension fund membership has been greater in themanufacturing sector than the service sector, and much greater in large companies than small and medium-sizedcorporations. Indeed, pension fund membership tends to be relatively greater in large companies than smallercompanies in many advanced economies.

20Association Générale des Institutions de Retraites des Cadres (AGIRC) and Association des Régimes de RetraitesComplémentaires (ARRCO).

1999, and by 2003 15 percent of the eligiblepopulation had enrolled in DC plans.

In many countries, the newer designs gener-ally are intended to develop multi-pillarfunded schemes, to supplement Pillar 1 as thetraditional primary source of retirementincome. These reforms include major changes(often reductions) in Pillar 1 programs,expanded funded corporate schemes (gener-ally DC or hybrid), and the development ofindividual retirement savings vehicles.21

Germany, for example, is moving towardfunded hybrid pension schemes.22 The exist-ing Pillar 2 schemes are primarily DB plans.Among them, Direktzusage (or “book reserve”—historically the most popular DB scheme withlarge German corporates) has not beenfunded by segregated assets, but the pensionfund liabilities are included directly in thecompany balance sheet, backed by the oper-ating assets of the sponsor company andconsidered “internally funded.” The rangeof occupational pension schemes has beenexpanded in 2001 with the creation ofPensionsfonds, which can be set up as eitherDB or DC schemes. Furthermore, “hybrid”schemes (in Pillar 2 and in Pillar 3) are grow-ing in Germany, many of which provide prin-cipal protection and minimum guaranteedreturns on accrued contributions in order toqualify for favorable tax treatment. Employeestypically are required to take an annuity onretirement.

Italy and France, through Fondi Pensione andPlan d’Epargne Retraite Collectifs (PERCO), haveestablished pure DC schemes for all privatesector employees (and public sector employ-ees in Italy).23 In both countries, these DCschemes are required to offer participants amenu of investment options with differentrisk-return profiles. As in Germany, the tax

regime in Italy encourages the payment ofbenefits at retirement through annuities,rather than as a lump sum.

Key Influences on Pension Funds’Financial Management

In addition to the challenge of aging popu-lations, a number of other factors influencethe management of pension funds. Nationalfinancial market characteristics, regulationsand tax policy, pension guarantee schemes,and accounting standards have a significanteffect on asset allocation and risk manage-ment strategies.

Financial Market Characteristics

As discussed in the April 2004 GFSR studyon the life insurance industry, national marketcharacteristics play a significant role in influ-encing institutional investment styles and pref-erences. Pension funds, like otherinstitutional investors, show a high degree ofhome bias in their investment strategies. Assuch, national markets may supply or limit theinvestment alternatives desired by pensionfunds to meet their specific investment needs.

However, pension fund investment behaviorcan be quite different from other institutionalinvestors in the same country or region, sug-gesting that regulatory and other factors arealso influential. In markets with relativelydeveloped funded DB plans, equities form alarge part of pension funds’ aggregate invest-ments. In some countries, this contrastssharply with the life insurance industry. Forexample, in the United States, pension fundsare much more heavily invested in equitiesthan insurers, despite the large domestic avail-ability of corporate bonds and other credit

KEY INFLUENCES ON PENSION FUNDS’ FINANCIAL MANAGEMENT

97

21See, for example, Allianz Dresdner Asset Management (2003).22Changes in German pension schemes are taking place in the context of the “Riester Reform” (2000–02).23Loi Fillon (2003) in France and the Berlusconi measures of 2003–04 in Italy. The recent Italian measures fol-

lowed the 1992–93 d’Amato reforms and the Dini-Prodi reforms of 1995–97 that introduced one of the most radi-cal pension reforms across industrial countries, switching from a PAYG DB scheme to a DC system.

instruments. Similarly, in European countriespension funds have not followed insurers,which have increased corporate credit invest-ments following recent pressure on solvencymargins and improvements in risk manage-ment techniques. This suggests that the lackof risk-based incentives, implemented perhapsthrough funding requirements or pensioninsurance premiums, or the relative sophisti-cation and adoption of risk managementtechniques may be at least as important adeterminant of investment strategies as thecharacteristics of local or regional capitalmarkets.

An important issue for pension funds is theavailability of long-term and index-linkedbonds. As routinely stressed by pension fundmanagers, financial products such as annuitiesand long-dated and index-linked debt instru-ments may better match pension liabilitieswith an average duration often beyond 20years, as well as addressing the needs of indi-viduals for Pillar 3 savings products. The mar-ket for long-term bonds is deepest in theUnited States (although, even there, the sizeof the market for maturities beyond 10 years isrelatively modest). A number of countries—for instance, France, the United Kingdom,and the United States—have small but grow-ing markets for index-linked bonds (see Table3.5). The United States has recently widenedits maturity range of issues to include a 20-year Treasury Inflation Protected Security(TIPS) bond, and Germany and Switzerlandhave also announced their intentions to issuetheir first inflation-linked bonds in 2005. Butin all mature markets such long-term instru-ments remain small compared with the size ofpension fund portfolios.

As a result, pension funds have soughtother ways to increase or match duration, and

some have turned to equities for long-termhedges. Given supply constraints on long-termor index-linked bonds, some pension fundshave relied on equities or other instrumentsto provide more duration or inflation hedges.This explains the relative significance ofequity holdings or real estate in many pensionfund portfolios. In addition, derivative instru-ments (such as swaps) have attracted somepension fund managers seeking to increaseasset duration or obtain some form of infla-tion protection.

Certain policy actions may be needed tostimulate further issuance of long-term andindex-linked bonds and to support these mar-kets. The availability and development of suchinstruments should be supported by nationalgovernments, including through governmentissuance and clear and consistent tax policyregarding long-term bonds. This shouldenhance pension funds’ ability to act as long-term providers of capital and support finan-cial stability. Corporate issuers desiringlong-dated funding exist in most mature mar-kets, such as capital-intensive industries, utili-ties, financial services (banks and insurers),and housing. In some areas (e.g., Europe),the development of securitization and struc-tured credit markets may also provide suchinstruments. Insurance companies willundoubtedly have an important role to play inthe expansion of annuity markets; however,even here, insurers need long-term marketinstruments to efficiently hedge and priceannuity risk.24 One particular factor that mayinhibit the supply of annuities by insurers maybe the difficulty of managing longevity risk ofthe extreme elderly as average life expectancycontinues to rise (often by more than earlierprojections). Backstop government funding ofthis “tail risk” could be an option to consider


98

24Annuities may provide payments either for the lifetime of the beneficiary or for a fixed term. In the UnitedKingdom, DC pension funds are required to provide 75 percent of pensions via lifetime annuities, and increasinglyfewer insurers are willing to sell such products. In the United States, by contrast, most annuities are fixed-term,thus presenting fewer hedging challenges. For a broader discussion of the challenges in developing such markets,see Jackson, Perraudin, and Trivedi (forthcoming).


99

Aging is expected to have far-reaching impli-cations for the global distribution of growth,labor, and capital. While aging is a global trend,there are large differences in its speed acrosscountries and regions. These differences, com-bined with the reforms of numerous nationalpension systems, may have a significant impacton the overall supply of capital, the perform-ance of capital markets, and international capi-tal flows.

Life Cycle and Supply of Capital

According to the traditional life-cycle theoryof consumption and savings, national savingsrates are expected to decrease in an aging econ-omy. To make up for lower income duringretirement, individuals would save an increasingfraction of their income during their workinglife and dissave during retirement. This wouldresult in a hump-shaped savings profile over aperson’s life (see the Figure).

Recent reforms toward multi-pillar pensionsystems are likely to validate this theory. Forinstance, in European countries with predomi-nantly public PAYG systems (e.g., France,Germany, and Italy), no old age dissavings hasbeen observed (indeed, intergenerational trans-fer of savings to younger relatives is takingplace), whereas in the Netherlands and theUnited States, where a large share of retirementincome is provided through private pensionschemes, the hump-shaped life-cycle savings pro-file is evident. This suggests that reforms towardmore balanced multi-pillar pension systems mayinduce both increased savings for retirementamong European workers and a decline in sav-ings rates at or near retirement.

The Potential Benefits of International Diversification

Investing pension assets internationally maybe good not only for risk diversification but alsoto realize better returns. With the growth offunded pension plans and the removal in somecases of investment restrictions, increased atten-tion has been paid to the international invest-ment of retirement savings. Overlappinggenerations models applied to advanced

economies show that substantially higher aggre-gate savings rates can be expected in an openeconomy than in a closed economy. In a closedindustrialized economy, an increase in nationalsavings leads to a larger capital stock and to adecrease in the rate of return on capital, whichacts to crowd out additional savings. In an openindustrialized economy, more savings are gener-ated as the rate of return does not change sig-nificantly. Indeed, research indicates that notonly a country or region’s absolute age struc-ture (and thus capital supply) but relative dif-ferences in age structure across countries orregions are an important determinant of capitalflows.1

However, the degree to which relative agingmay determine capital flows will also depend onthe international mobility of capital. In thisregard, existing frictions, such as taxation or for-eign investment limitations, together withinvestors “home bias,” may limit the benefits ofinternational diversification.

1See for example Higgins (1998), Reisen (2000),and Lührmann (2002).

Box 3.2. Individuals’ Life-Cycle Savings and Global Capital Markets

–10

–5

0

5

10

15

20

25

30

75+65-7455-6445-5435-4425-34

Estimated Personal Savings Rates for Selected Countries by Age1

(In percent)

Sources: Schieber (2003); and Börsch-Supan (2004).1Ages are of the household head/reference person; and savings

rate in each age group is for all households in their respective samples.

Japan

Italy

Germany

United States

NetherlandsCanada

United Kingdom

if it avoids a greater Pillar 1 cost arising from ashortage of supply of annuities.

International diversification can help over-come national or regional market constraintsand support macroeconomic savings patterns.National savings rates are likely to declinein aging societies, due to life-cycle effects.This may encourage greater investment ofpension assets into the economies of younger,economically faster-growing countries ormarkets. Indeed, there is evidence that suchinternational diversification not only pro-vides benefits from risk diversification, butmay also provide higher returns on capital(Box 3.2). This highlights the importanceof pursuing regulatory efforts to eliminatedomestic investment restrictions and promoteimprovements in risk management at pensionfunds.25

Taxation and Regulation

Regulatory and tax constraints on invest-ment behavior and national funding rules sig-nificantly influence pension fund strategies.Among the rules set by a variety of bodies, taxrules tend to have the greatest influence onannual funding decisions by pension sponsorsand on individuals with regard to retirementsavings.

Taxation

In many cases, tax rules on pension contri-butions effectively set upper and lower boundsfor funding decisions. This is the case in theUnited States, where contributions that wouldincrease the funding level beyond a 100 per-cent funding ratio are not tax deductible, andeven attract an additional 10 percent excisetax. At the same time, tax policies and penal-ties also aim to prevent large funding deficien-cies, such as through the imposition of a taxof 100 percent of the deficiency in case of fail-ure to correct it.

Taxation and other rules can create disin-centives or prohibitions to annual contribu-tions or the withdrawal of surplus assets,thereby further discouraging precautionaryoverfunding. In the United States, excessassets cannot be withdrawn by companiesfrom pension schemes unless the scheme isterminated, and then up to a 50 percent dutyplus standard corporation tax would need tobe paid. Meanwhile, as noted above, the lossof annual deductibility and a 10 percentexcise tax on contributions to overfundedschemes were driving factors behind the con-tribution holidays in the 1990s, and increasethe potential risk of schemes becoming under-funded in the event of adverse market orother developments. Allowing moredeductible (or at least not penalized) contri-butions, up to some reasonable overfundinglimit (e.g., two or three years of “normal” con-tribution rates) would give sponsors and pen-sion managers a greater ability to prudentlyplan for cyclical downturns.

Tax rules for savings products play a strongrole in determining how individuals save (Box3.3). Tax rules are generally designed to givepreferential treatment to retirement savings,and as an incentive for individuals to start sav-ing early. Indeed, if private savings are insuffi-cient to ensure adequate old-age income inthe long run, the cost of providing a safety netwill ultimately fall on the government (viaPillar 1). As such, the fiscal cost of tax incen-tives for retirement savings may be viewed aspreventative of potentially much larger coststhat may be incurred later to support personswith inadequate pension savings.

However, it remains unclear whether taxincentives help raise overall pension savings,or merely shift existing savings. In particular,complex taxation regimes favoring certaintypes of pension plans may simply reallocatesavings with little or no increase in the overallsavings level. Empirical evidence is mixed, but


100

25See, for example, Schieber (2003) and Deutsche Bank Research (2003).


101

The tax treatment of pension plans in most OECDcountries is broadly as follows (see the Table):• Contributions by employees into approved pension

schemes are deductible from the employees’ tax-able gross income (i.e., they are made before tax),and contributions by employers are deductiblefrom the employer’s earnings. In general, thetotal deductible contributions are limited to amaximum percentage of the employee’s income.

• Income earned by approved pension funds fromtheir investments is exempt from tax.

• Pension income received by individuals is normallysubject to tax on the same basis as wages, andearly distributions or distributions less than aminimum required level are both generally sub-ject to additional taxes.Tax systems usually treat qualified pension plans

preferentially, with the net fiscal cost of these incen-tives varying widely in the countries shown in theTable. The fiscal cost is measured as the difference,over the length of the investment, between theamount of taxes collected when (a) the money issaved in a pension plan and (b) if it were invested ina benchmark non-retirement saving vehicle.Estimates for 2000 range from 1.7 percent of GDPin the United Kingdom and 1.5 percent inSwitzerland, to 1.0 percent in the United States andthe Netherlands, and 0.2 percent in Japan. However,the revenue forgone from tax incentives for privatepensions remains a small fraction of the govern-ments’ spending on public pensions. In Germany,spending on public pensions was about 100 timeslarger than the forgone revenue from pension sav-ings in 1997 (0.1 percent of GDP), while it was eighttimes in Japan, seven times in the Netherlands, sixtimes in the United States, and three times in theUnited Kingdom.

However, poorly designed tax systems may resultin simply substituting one form of savings foranother, with little or no additional pension savings,and larger deadweight losses. Complex taxationregimes can generate distortions in favor of one typeof pension plan versus another, or in favor of pen-sion plans relative to other saving vehicles. Therehave been moves to simplify tax regimes and, insome countries, grant a single tax treatment for alltypes of occupational pension schemes. In France,

pension reforms have set up a new legal frameworkfor all private retirement savings (Pillars 2 and 3),consisting of an annual global tax deduction. Thesame global approach, encompassing Pillars 2 and 3,has been introduced in Germany, but the complex-ity of tax incentives and savings subsidies is inhibit-ing the take-up of retirement savings products.

While an important goal of tax incentives target-ing occupational pension funds and other retire-ment saving is to raise the level of national savings,empirical evidence in this regard is mixed. In gen-eral, studies have found a minimal impact of occu-pational pension plans on national savings.However, many methodological issues affect the reli-ability of the results. A recent survey (OECD, 2004b)finds that about 60 to 75 percent of savings in tax-favored pension vehicles simply displaces other sav-ings. Some studies also indicate participation ratesin pension schemes are affected by taxes. A U.S.study (Reagan and Turner, 2000) found that a 1 per-centage point increase in marginal tax rates leads toa 0.4 percentage point increase in the proportion offull-time employees participating in a pension plan.

Box 3.3. The Tax Treatment of Pension Plans: a Comparison for Selected Industrial Countries

Tax Treatment of Pension Plans

Pension Benefits_______________Fund Lump

Countries Contributions Income Annuities sum

France T/PE1 E T/PE T/PEGermany T/PE E T T/PE2

Italy T/PE3 E T/PE4 T/PE5

Japan E E T/PE T/PENetherlands E E T TSwitzerland E E T TUnited Kingdom T/PE E T T/PE6

United States T/E7 E T/E8 T/E8

Source: OECD (2004b).Legend: T = taxed; E = exempt or deductible; PE = partially exempt

or deductible. 1Deductible up to 19 percent for up to 8 times the annual social

security ceiling (€44,360 in 2003).2Tax-free allowance of 40 percent of pension payments granted up

to €3,072 at age 63 or higher.3Tax exempt up to 2 percent of gross employee earnings.4Taxed only for 87.5 percent of their gross amount.5Taxable base limited to the part over the employee’s contribution

to the fund.6Tax-free lump sum of up to 25 percent of fund value.7Exemption up to $12,000 for 401(k) plans (employees) and up to 3

percent of employee compensation (employers matching dollar fordollar contributions). Contributions to Roth IRA are not tax deductible.

8Income from Roth IRA is tax exempt.

a recent survey (OECD, 2004b) finds thatabout 60 to 75 percent of savings in tax-favored vehicles represent a reallocation fromother savings (see Box 3.3).26 Even such areallocation of savings, however, may be bene-ficial in encouraging retirement planning if itrepresents a shift from short-term to longer-term and more stable savings.

Regulation

Supervision of pension funds traditionallyhas been conducted by bodies primarily con-cerned with labor and benefits, rather thanfinancial markets. Thus, to date much of pen-sion fund regulation has focused on the pro-tection of pensioner and employee rights, andensuring that pension fund assets are segre-gated for the benefit of employees, ratherthan reviewing the risks and long-term dyna-mic process of assessing whether the obliga-tions will be met. Nevertheless, pensionregulators often set minimum fundingrequirements and, in some cases, restrict cer-tain investments or asset holdings, and thusinfluence investment behavior.

The choice of the discount rate for mini-mum funding requirements heavily influencespension fund asset allocation strategies.27

Pension fund managers wishing to limit thevolatility of their regulatory funding ratio mayhold a larger allocation of assets with a highcorrelation to the discount rate used for liabil-ities. Corporate bond yields are increasinglyused by regulators as the discount rate for lia-bilities, and this should increase pensionfunds’ demand for credit instruments. In theUnited Kingdom, discount rates based oninflation-linked yields stimulated growingdemand for such products. In the late 1990s,a shift to government yields, at a time ofshrinking government debt supply, led to verylow (and at times quite volatile) yields on

these instruments. The more recent move toAA corporate bond yields provides a widerrange of potential issuers for investment andhedging purposes, and removed the regulato-rily driven pressure on government yields.

In the United States, the discount rate forfunding calculations has been temporarilyamended to a corporate bond rate, in aneffort to provide short-term relief to under-funded plans. For two years, the discount rateused to determine DB scheme liabilities andsponsor companies’ required contributions(as well as their PBGC premiums) will be afour-year weighted average of long-term high-grade corporate bond yields, replacing the30-year U.S. treasury bond yield. In addition,certain industries (for example, steel andairlines) benefit from specific financialsupport—for instance, only 20 percent ofannual contributions that would be otherwiserequired to address underfunded situationsare to be contributed each year, thus spread-ing out the required increased payments—(Federal Deposit Insurance Corporation,2004).

Regulations also influence asset allocationthrough “prudent person” rules and formallimits on certain investments. In many coun-tries, regulators explicitly restrict the range ofinvestment options by imposing quantitativeinvestment limits, usually by asset class (Yermo,2003). Although some countries continue toplace upper limits, for instance, on invest-ments in foreign securities, regulatory con-straints on pension fund allocations rarely actas a major constraint on investments today.28

The “prudent person” rules generally establisha principle of “diligence that a prudent per-son acting in a like capacity would use” (Galer,2002). Fear of liability under those rules canlead fund managers to invest in portfolios thatare substantially similar to their peers, and


102

26OECD (2004b).27See, for example, Blake (2001).28In the past (as described earlier), Japanese restrictions on allocations both to equities and to foreign securities

acted as a strong constraint on pension fund allocation strategies, but these were abolished in 1997.

therefore can constrain pension funds fromdeveloping innovative or new approaches, andquite possibly from developing more modernrisk management approaches. It can alsoinduce herd behavior, and thereby introducemore volatility to capital markets.

Historically, many industry observersdescribed pension fund regulation as unso-phisticated in dealing with solvency and riskmanagement, but there are signs of change.The regulatory factors mentioned above donot deal explicitly with the risk of investmentportfolios. They often focus, in a more limitedway, on the current level of funding, and arebased on a variety of qualitative assumptionsabout future performance, and not on therisks inherent in the pension fund’s asset-liability mix. However, some regulators arebeginning to take a more sophisticatedapproach to evaluating the risk profile of pen-sions. In the Netherlands, a combined regula-tor has been established for insurers andpension funds, the Pensioen & Verzekeringskamer(PVK), and its merger with the banking regu-lator is expected to be formally completed inJanuary 2005. PVK is importing many of therisk principles and measures applied to finan-cial institutions into its pension supervision(Box 3.4). In addition, pension guaranteefunds, like the PBGC or the PensionProtection Fund (PPF) being developed inthe United Kingdom, are considering takingaccount of portfolio risks in the premiumsthey establish for individual pension plans.The developments essentially act to introducerisk-based capital or funding requirements tothe pension system.

Initial steps have been proposed to establishinternational minimum standards for pensionregulation. The OECD (2004a) recentlyissued Core Principles of Occupational PensionRegulation. While rather general in scope, theyhave proposed principles relating to, amongother things, full funding of pension schemes

and the enhancement of portability. Weencourage further progress to apply anddevelop principles such as these.

Pension Guarantee Funds

The social objective of encouraging andprotecting private pension savings has also ledto the creation of pension guarantee or insur-ance funds. Guarantee funds are intended todiversify the risk of pension fund failuresamong the general population of pensionplans, and should eliminate or (at least)reduce the potential cost to the government,if it were to act as the ultimate safety net forpensions. Guarantee funds are likely toincrease in importance, as more countrieslook to increase the role of private pensions.The United States (PBGC), Germany (PensionsSicherungs Verein), and Switzerland (theGuarantee Fund) have long-standing insur-ance funds, and the United Kingdom is look-ing to establish such a fund (PPF).

However, this insurance protection may cre-ate other risks, depending on how guaranteefunds are designed or operate. Most impor-tantly, guarantee funds may generate moralhazard, to the extent they lead weaker spon-sors to increase investment risk in the pensionfund in the hope of reducing or limiting con-tributions. In the United States, for example,PBGC “risk-based” premiums relate only tothe degree of underfunding and do not takeinto account the asset mix or liability struc-ture. In addition, if a guarantee fund’s owninvestment portfolio tends to have a similarasset mix as that of the covered pensionfunds, then the guarantee fund may be expe-riencing difficulties when claims from dis-tressed pension funds are greatest. This maybe exacerbated if pension funds tend tobecome underfunded when sponsor compa-nies face more difficult business conditions(i.e., cyclical).29


103

29These challenges are illustrated by the PBGC, whose deficit reached a trough of over $11 billion at the end ofits 2003 financial year, after having a surplus of $10 billion in 2000.


104

A proposed redesign of pension fund supervisionin the Netherlands aims to ensure that pensionfunds remain fully funded at almost all times. Strictrules are being proposed for the rebuilding offunding levels in pre-specified timescales, either byincreasing funding or reducing the indexation ofbenefits. Pursuant to the proposal, three parallelfunding tests would be applied:• a minimum test, requiring pension funds to main-

tain a minimum 105 percent funding ratio, evenif their assets and liabilities are perfectlymatched;

• a continuity test, requiring pension funds with con-ditional indexation clauses to have a long-termplan to meet their conditional goals during thenext 15 years; and

• a solvency test, reflecting the composition of thefund’s assets.The solvency test, in particular, would introduce

an innovative risk-based capital framework for pen-sion funds. The risk parameters would be set so as toguarantee at a 97.5 percent confidence level that thefunding ratio will stay above 105 percent over oneyear (taking account of expected contributions andexpenses during the year). In other words, the fund-ing ratio for funds meeting this requirement wouldonly be expected to fall below 105 percent onceevery 40 years.1 If the funding ratio fell below therisk-based floor, the fund would be granted a periodof 15 years to address this gap (either throughincreased contributions or reduced investment risk).This risk component is meant to provide, like BaselII, a standard risk measure set by the supervisor, oralternatively allow funds (where appropriate) to usetheir own risk models and capital calculations.2

For the standard calculation, assets would bemarked-to-market, and liabilities measured with a“market” yield curve. The discount rate would beset according to the duration of liabilities, and may

reflect a government yield curve. The volatilityparameters for assets would be based on a long-termhistorical run of data, reflecting the long-run orien-tation of pension funds. Liability measures also areexpected to assume further increases in longevity(e.g., a two-year lengthening of average life spans).

Companies would also be required to statewhether they have a “conditional” or “uncondi-tional” inflation-indexation policy for pensions. Ifthe policy is unconditional, the 105 percent regula-tory floor would need to be against inflation-linkedliabilities. However, if the policy is conditional, liabil-ities need only be measured in nominal terms. Sincemost pension funds seem to have opted for this con-ditional form, benefit commitments would generallybe assured in nominal terms, and indexation wouldbe contingent on investment performance or a com-pany’s willingness (but not legal commitment) toincrease contributions. This requirement is viewedas a means to communicate the risk and protectionsprovided to pension beneficiaries.

Overall, the proposed rules will give pensionfunds multiple hedging goals. Such a regulatoryframework will lead pension funds to view asset-liability management (ALM) as an exercise inhedging both nominal liabilities (to meet theirsupervisory funding floor) and, possibly, real liabili-ties (to meet a conditional indexation goal, ifretained). Which of these aims is more importantwill depend on their funding position. Weaker pen-sion funds may begin hedging with bonds and aban-don an indexation goal, while stronger funds,operating above their risk-based capital floor, maycontinue to target higher real returns, using greateramounts of equities and index-linked bonds, whereavailable. However, many funds are expected to pur-sue a mixed approach, holding nominal bonds tomeet their 105 percent liability floor, and investingthe surplus in riskier assets in order to possiblyachieve indexation goals. While Dutch pensionfunds currently tend to hold diversified portfolios(with a typical portfolio consisting of 50 percentequities, 40 percent bonds and loans, and 10 per-cent real estate and other investments), some fundshave already reduced their equity allocations andsought to increase the duration of fixed-incomeassets to meet the proposed supervisory framework.However, to date such portfolio changes have notbeen widespread.

Box 3.4. Proposed Risk-Based Capital System for Pension Funds in the Netherlands

1Although the exact parameters have yet to be estab-lished, the supervisor estimates that a fund invested50/50 in bonds and equities, and with a typical bondduration profile of five years, could be expected tohave a minimum risk-based capital requirement of 130percent of projected liabilities.

2The supervisor expects about 10 to 20 of the largestpension funds to apply internal models, and others touse the standard measures.

We support the inclusion of more risk-basedelements in the design of guarantee funds.Risk-based premiums are being considered inthe United Kingdom that may (at a mini-mum) take account of the investment or mar-ket risk in pension fund portfolios. Moregenerally, risk-based premiums could be basedon various criteria, including funding levels(based on accumulated benefit obligations, orABOs, which seem particularly appropriate foran insurance fund), asset composition, liabil-ity structure (e.g., average maturity or dura-tion), and degree of asset/liability matchingof the pension fund.

Finally, whether and how guarantee fundsor regulators should take account of thesponsor company’s financial strength remainsan open question. In principle, the cash flowand balance sheet strength of the sponsorcompany should play a role in determiningthe pension fund’s ability to meet its liabili-ties. However, in practice, the great diversityof companies across a wide range of indus-tries would make the evaluation of theirfinancial strength an even more difficult taskfor supervisors than they currently face forsingle industries, such as banking or insur-ance. Moreover, the deterioration of a pen-sion plan’s funding level and/or an increasein its holding of “risky” assets may reflect itsown ability (or not) to support the pensionfund, and therefore these criteria may satisfythe supervisory need to set objective risk-based premiums.

Accounting

Accounting is frequently cited as the mostimportant factor affecting pension fund man-agement, and the shift from DB to DC orhybrid schemes.30 Pension obligations canintroduce volatility in the sponsor company’sfinancial statements, depending on how they

are measured and recorded. Indeed, industryobservers frequently assess that a move to mar-ket-based, fair value accounting principleswould significantly increase the shift awayfrom DB pension plans and may encouragegreater short-term trading and investmentstyles.

Current Practices

In most jurisdictions, the impact of short-term pension gains and losses on the financialaccounts of sponsor companies are smoothedover several periods. Historically, a variety ofsmoothing practices have been applied to vari-ous components of a pension sponsor’s finan-cial statements, including investment returns(actual against expected), and actuarial gainsand losses (i.e., changes in liability values).The current international accounting stan-dard (IAS 19) and national accounting stan-dards in most of continental Europe, Japan,and the United States incorporate varioussmoothing mechanisms (see Box 3.5).

Another important accounting principle isthe choice of the discount rate used to meas-ure pension liabilities.31 This rate has a signifi-cant influence on the measurement of theobligation, as a higher rate reduces the pres-ent value of pension obligations. Indeed,some analysts have suggested that the rateselected or movements in rates have a greaterinfluence on pension fund balance sheetsthan asset performance, given the typicallylong average duration of liabilities. Some juris-dictions have allowed the same discount rateto be used for liabilities as for expectedreturns on assets, thus further smoothing theimpact of market movements (such as the pro-jected yield on equities). However, in accor-dance with IAS 19, many jurisdictions nowrequire a rate approximating a high-quality(AA or equivalent) corporate bond yield. Inother countries, like Germany, the discount


105

30See, for example, CIEBA (2004).31The discount rate used in the financial accounts is not always the same as the discount rate used for regulatory

purposes.


106

Approaches to pension accounting differ sig-nificantly across countries. The differenceslargely relate to the degree to which theaccounting permits smoothing in considerationof uncertainties associated with pension-relatedcosts and obligations, the subjective and com-plex process of estimating the obligations, andthe long-term nature of the obligation. This boxcompares three pension accounting regimes,namely, U.S. FAS 87, U.K. FRS 17 (to be fullyimplemented in January 2005), and proposedIAS requirements (IAS 19, effective January2005).

How Pension Assets and Obligations Are Measuredand Presented in Corporate Balance Sheets

Under all three regimes, the sponsor com-pany recognizes pension obligations net of pen-sion assets. However, in measuring pensionassets and liabilities, the U.S. regime (FAS 87)allows more smoothing than FRS 17. Both IAS19 and FAS 87 permit amortization of unrecog-nized gains or losses over the remaining work-ing life of active employees, but also permitmore rapid, and even immediate, recognition.Under IAS 19 and FRS 17, pension assets aremeasured by market values. Under FAS 87, pen-sion assets are measured at either market valueor a calculated value that recognizes changes infair value over not more than five years(referred to as “market-related value”). Underall three regimes, liabilities are measured bythe projected benefit obligation (PBO). PBOmeasures obligations on the assumption thatthe plan remains a going concern, and so ismeant to capture the impact of future wageincreases and unvested benefits, actuarialassumptions, and discount rates determined asof the current measurement date. Pension lia-bilities, under FAS 87, are measured based onthe PBO with a requirement to recognize anadditional minimum liability if the accumulatedbenefit obligation (ABO), which representsessentially a liquidation value, exceeds the fairvalue of plan assets; both the ABO and PBOamounts are disclosed in the notes to the finan-cial statements.

If there is a net surplus in the pension fund,the sponsor company may record all or part of itas an asset. While FAS 87 sets no explicit limiton the amount that may be recognized, IAS 19and U.K. FRS 17 limit it to the amount thatwould be recoverable by the sponsor through arefund or a reduction of future contributions.

Regarding the discount rate to be applied topension liabilities, IAS 19 prescribes yields ofhigh-quality corporate bonds; U.S. FAS 87 givesa choice of either high-quality corporate bondsor insurance annuity rates; and U.K. FRS 17 rec-ommends AA or equivalent corporate bondyields.

Smoothing Principles in the Profit and Loss Account

In general, when evaluating pension fundinvestment results, sponsors may take a long-term view by smoothing short-term performancevolatility. For this purpose, IAS 19 and U.S. FAS87 reflect expected returns rather than actualreturns on pension assets. The differencebetween actual and expected returns is subjectto amortization in future periods, or at timesmay be entirely deferred if it does not exceed aminimum threshold. The rate of expectedreturn reflects each company’s view about thefuture performance of its pension portfolio.

Under IAS 19 and U.S. FAS 87, smoothingalso exists in actuarial gains and losses (i.e., pro-jected liabilities), which are also amortized andreflected in earnings over future periods. If thedifference between actual and expected returns,together with other actuarial gains or losses, iswithin a range of 10 percent of the higher ofplan assets or liabilities (the “corridor”), theamount is not required to be amortized. UnderIAS 19 and FAS 87, plan sponsors may elect asystematic method of amortization that must beapplied consistently (see the Table).

The United Kingdom’s FRS 17 also usesexpected returns; however, the differencesbetween expected and actual returns, as well asactuarial gains and losses, are recognized in theperiod in which they are incurred in a separateStatement of Total Recognized Gains and Losses(STRGL). Use of the separate account, instead

Box 3.5. Comparison of U.S. FAS 87, U.K. FRS 17, and Proposed IAS Standards

rate is fixed by the authorities and only rarelyadjusted.

Recent Trends

The trend among standard setters is towardlimiting the scope for pension fund smooth-ing, by introducing more market sensitive or

fair value principles. The United Kingdom ismoving toward a fair value approach with theintroduction (to be completed in 2005) of anew accounting rule (FRS 17). Under thisrule, although the “headline” profit and lossaccount continues to show the actuarialversion of pension gains and losses, the


107

of direct reporting in the profit-and-loss state-ment, is an attempt to avoid introducing exces-sive volatility into headline income figures.Recently, the IASB has issued a proposal on thepossible introduction of a separate account to

allow companies to report the annual cost (withor without smoothing). It should be noted thatIASB and U.S. FAS 87 also allow immediaterecognition of the difference between actualand expected returns at the company’s option.

Important Differences in the Three Accounting Standards

IAS 19 U.S. FAS 87 U.K. FRS 17

Measurement of Projected Benefit Obligations PBO. PBO. pension obligations (PBO). Accumulated Benefit Obligations

(ABO) (minimum recognition):PBO and ABO are reported in the notes to the financial statements.

Measurement of Fair market value: no smoothing Market-related value: companies Fair market value: no pension plan assets allowed. are permitted to use fair market smoothing allowed.

value or a calculated value that smoothes up to five years for purposes of determining the asset value for use in the return on assets and 10 percent corridor computation. The value of assets disclosed in the notes is the fair market value.

Smoothing of gains Unamortized past service costs Unamortized past service costs Unamortized past service costsor losses in earnings are amortized over the remaining are amortized over the remaining are amortized over the period statements service period. service period. in which the benefits vest.

Actuarial gains or losses within a Actuarial gains or losses within a The difference between“corridor” may be ignored (the “corridor” may be ignored (the actuarial gains, losses, and higher of 10 percent of the present higher of 10 percent of the adjustments is recognized in value of the obligation or 10 percent present value of the obligation or the period incurred in aof the market value of assets). 10 percent of the market- separate note in the financial Actual gains or losses over a related value assets). statement (STRGL), i.e., not “corridor” may be amortized over Actual gains or losses over a smoothed.the remaining working life of “corridor” may be amortized. Theactive employees (immediate minimum required amortizationrecognition is permitted). is based on the remaining working

life of active employees.

How future investment Long-term estimates of expected Long-term estimates of expected Long-term estimates of returns are calculated returns. returns. expected returns. However, the

difference between expectedand actual returns is recordedin STRGL.

Sources: Standard & Poor’s (2003); and Financial Accounting Standards Board.

unsmoothed mark-to-market version of thegains and losses are shown in a separateStatement of Total Recognized Gains and Losses(Box 3.5). The International AccountingStandards Board (IASB) has also introducedchanges in its pension accounting standardsthat will permit reporting according to fairvalue principles in a form similar to theUnited Kingdom—EU countries agreed toadopt IAS 19 in January 2005.

In many jurisdictions, steps are also beingtaken to ensure greater disclosure of a pen-sion fund’s financial condition. Pension liabili-ties are increasingly reflected like other debtobligations of the sponsor company. Japanbegan recording pension liabilities as debtobligations of the sponsor in 2000—previouslyJapanese companies were required only torecognize annual contributions as an expensein the profit and loss account. This move hasforced many small and medium-sized enter-prises to terminate their pension plans due tothe sudden reporting of large funding gaps intheir balance sheets.

Potential Impact

The use of fair value accounting principleswould address the arbitrariness that character-izes traditional pension fund accounting prac-tices. It is widely recognized that the varioussmoothing mechanisms used in the account-ing for pension plans introduce an arbitraryand inconsistent application of currentaccounting standards, which some argue sub-stantially limits the usefulness of financialreports. In particular, the use of subjectiveassumptions, which frequently vary betweencompanies, may hamper comparative analysis,and the financial risks borne by the sponsorcompanies may be underestimated (Shilling,2003).

However, it is also argued that by generat-ing greater volatility in sponsor companies’

balance sheets, fair value accounting princi-ples may misrepresent (i.e., over- and under-state) a pension fund’s financial conditionand accelerate the shift away from DB plans.Recent experience in the United Kingdomindicates that fair value principles may accel-erate moves to DC and hybrid plans, whichallow companies to reduce their risk concern-ing pension obligations and transfer invest-ment and market volatility to employees/beneficiaries. Similar effects can be seen inJapan and the United States. Greater sensitiv-ity to market price volatility may also in thefuture encourage fund managers to focus onshort-term asset management strategies, oralternatively to seek to immunize themselvesfrom short-term accounting volatility by reallo-cating their portfolios from equities to bonds.

Rating Agencies

Rating agencies now explicitly recognize theunderfunded amount of pension plans asdebt of the sponsor company. The ratingagencies treat the difference between the PBOand the fair value of plan assets like any otherlong-term debt obligation of the sponsor com-pany,32 and use various adjustors to unwindsome of the smoothing introduced by currentpension accounting practices.33 This shift inratings analysis has resulted in several ratingsdowngrades at least partly based on pensionissues, particularly in continental Europe.Such actions often affect companies in olderindustries, with an aging workforce and/or aperceived weaker cash-flow strength or finan-cial flexibility. Increased attention to the rat-ing impact of pension funding levels seemsalso a factor in the shift from DB plans to DCand hybrid schemes.

Recently, some rating agencies have startedto make explicit statements regarding pensioninvestment strategies, giving greater support


108

32By accepting a portion of their compensation on a deferred basis, employees essentially become creditors ofthe sponsor company.

33See, for example, Moody’s Investors Service (2003).

to fixed-income pension assets. Fixed-incomeassets are seen as providing greater security tobeneficiaries—at the expense of higherreturns. Most simply, based on this view, themore closely a pension’s projected obligationsare matched with a portfolio of high-qualityfixed-income securities, the greater its abilityto meet its liabilities as they fall due. Whilethis analysis seems sound, the ability toachieve such asset/liability matching is diffi-cult, and the availability of market securitiesmay be lacking.

Asset Allocation and Risk ManagementThe ultimate purpose of pension schemes is

to meet their committed future pension liabil-ities. The fund manager’s duty is to managethe fund for the benefit of the plan membersin order to meet those liabilities, rather thanto earn an excess return. Given the liabilities’generally long-term structure, this implies along-term focus to investment. A number ofrisks need to be managed as part of the ALMprocess, including the duration of both assetsand liabilities, inflation, longevity, and theability of the sponsor company to meet futurecontribution needs. Challenges and con-straints also arise, such as those concerningthe availability of appropriate financial instru-ments, the impact of pension fund perform-ance on the sponsor company’s accounts, andthe general desire to keep the level of contri-butions down.

Asset Allocation

There is no consensus among pension andinvestment experts on the appropriate assetallocation for DB or hybrid pension funds.Although there are many differentapproaches to investment management forpensions, asset allocation approaches gener-ally fall into one of three different styles.

Primarily Equity-Based

Many in the pension fund industry favor aportfolio consisting primarily of equities,largely because they believe that in the longrun the extra return from equities will out-weigh the short-term volatility. In their view,although equity returns can be volatile in theshort run, equities are much more likely overthe long-term average life of pension liabilitiesto outperform bonds, and thereby reducecontributions or allow for increased benefits.Accordingly, they also generally oppose “fairvalue” accounting methods, arguing that itdoes not reflect the long-term nature of pen-sions or pension investment.

Many advocates of this position also viewequities as a better inflation hedge than nomi-nal bonds, and that, given the lack of supplyof long-dated bonds, equities are a more prac-tical way to match the duration of pension lia-bilities. Equities are seen by some as a goodinflation hedge because their value reflectsfuture expected profits, and hence may beseen as likely to rise with future wage andprice growth in the long term. Some alsoargue that equities have a much longer dura-tion than bonds, because their dividends rep-resent a stream of cash flows with no finalmaturity or because their price movementscan be quite large in response to interest ratemovements.34 This would imply that theycould be useful as a hedge for long-term lia-bilities. On the other hand, other market ana-lysts find the correlation between equities andbonds is often weak, or too variable over time,to be relied upon as a duration hedge. Theequity market fall in 2000–02, which effec-tively implied negative duration of equities forthat period, was the largest two-year fall inmajor markets since the Great Depression,leading some to reduce their equity alloca-tions. Nevertheless, supporters of an equity-based strategy argue that the high returns ofthe 1990s outweigh the two years of losses.

ASSET ALLOCATION AND RISK MANAGEMENT

109

34For instance, Standard & Poor’s (2004b) have estimated a current duration of 15 years for the U.S. equity market.

Primarily Bond-Based

A recent body of opinion favors a portfoliobased wholly or primarily on fixed-incomesecurities. The argument is that, as a pensionfund’s liabilities form a future stream of pay-ment obligations that closely resemble a port-folio of bond payments, a bond portfolio canbest provide the certainty that the pensionfund will meet its liabilities as they fall due. Atthe same time, sponsor companies should notseek or accept additional business, leverage, orinvestment risks through their pension fund.Shareholders, it is argued, do not desire thisadditional market exposure. If shareholdersseek a diversified portfolio of this type, theycan more efficiently build one themselves.

Many companies indicated that they wouldconsider moving to a much larger bond allo-cation if their funding ratio rebounded to 100percent or more. The most publicized exam-ple of this strategy has been the U.K. retailfirm Boots, which moved to a 100 percentbond allocation in 2001. However, the com-pany has more recently announced that itintends to invest up to 15 percent in otherassets, to better match very long-dated liabili-ties, which extend beyond 35 years, and forwhich it is not possible to purchase equivalent-duration bonds. A few other employers (invarious countries) have also moved to a morebond-based investment strategy. However,many companies are reluctant to make signifi-cant short-term contributions or switch to cur-rently highly priced fixed-income instrumentsgiven their current weak funding levels.

A “Balanced” Portfolio, with Bonds, Equities, andOther Assets

Some pension funds, and their consultants,argue that a diversified investment portfoliocomposed of a variety of asset classes offersthe best way to balance risk and return.Pension fund managers supporting a balancedportfolio approach often also favor certain“alternative investments” (such as privateequity, real estate, commodities, and morerecently hedge funds) in addition to bonds

and equities (Greenwich Associates, 2003).Fund managers may employ the balancedapproach to seek to enhance return throughactive management of a variety of asset classes,while diversifying risk and perhaps matchingnear-term cash flows. Such an investment pol-icy has also been attractive to funds (forinstance in Switzerland or, historically, theNetherlands) that measured their liabilitieswith a relatively fixed discount rate and there-fore had a fixed asset return target, or man-aged the assets against benchmark indicesrather than against liabilities. The relativelysmall domestic markets in the Netherlandsand Switzerland have also led funds in thesejurisdictions to diversify internationally.

A vigorous debate is currently taking placein the pension fund industry on the merits ofthese different approaches. The debate onthese different strategies is also closely relatedto discussions regarding broader risk manage-ment, and accounting and regulatory issues.To illustrate some of the arguments regardingthe relative merits of bonds and equities, twoU.K. market analysts whom we met during thepreparation of this study agreed to provideshort pieces on their differing analyses. Theseare presented in Boxes 3.6 and 3.7.

Policymakers need not take a view on opti-mal asset allocation, but should ensure thatdecisions are guided by appropriate risk man-agement practices. Given the long-termnature of pension provision, some asset andliability risks certainly are being taken. Theserisks need to be understood and assessed byfund managers, and appropriate safety mar-gins encouraged through risk managementstrategies (e.g., a prudent level of overfund-ing). Policymakers can encourage thisthrough regulation and tax policy.

Risk Management

There is great variation in the sophisticationof pension fund management. Some of thelargest funds commit considerable staff andother resources to internal trading capacity,


110

risk analysis, and/or the management ofexternal fund managers. In other cases,medium-sized (and some larger) funds haveonly a handful of employees to evaluate bene-

fit obligations and determine asset allocation,and often have delegated much of thedetailed work to consultants and externalmanagers.


111

Shareholders Are the End Risk Bearers

Default scenarios aside, shareholders in a cor-poration bear the asset versus liability risk withinits pension plan (with gains or losses in assetsreflected in adjustments to the contributionrate). Shareholders should therefore be broadlyindifferent to the following three options: hold-ing equities (say) within the DB pension plan ofthe company they invest in; holding equities ofother firms directly on the balance sheet of thecorporation; or holding these equities in theshareholders’ personal portfolios. This analysisis similar in principle to the Modigliani andMiller (1958) indifference proposition.Furthermore, as with the Modigliani and Millerproposition, despite the apparent indifferenceat first sight, factors such as tax and frictionalcosts are the key to understanding optimal struc-tures in practice.

Arguments in Favor of Matching

Among the reasons why a shareholder shouldprefer the assets and liabilities of a pension planto be closely matched are:• The company reduces the likelihood that

financial losses in the pension plan disruptthe core business activity.

• The actions of management are more easilymonitored and the scope for internal cashwindfalls being lost or misallocated isreduced.

• Internal and third party management costsand fees are minimized.The cost of defined benefit obligations cannot

be reduced by investing in equities except in sofar as the value of the benefits to plan membersare reduced, by increasing default risk. Overall,

shareholders can only gain by this if they canavoid the policy rebounding in the form ofhigher wage costs, adverse publicity, or govern-ment responses (e.g., restrictions on corporateactivity).

Defined Benefit Liabilities Can Be Matched by UsingNominal or Inflation-Linked Bonds (or Swaps)

Debate over the close matching of DB liabili-ties often focuses unnecessarily on the link withfuture salaries. However, few would disagree thatthe accumulated benefit obligation (ABO) canbe very closely matched with bonds without tak-ing a view on future salaries, and, in fact, thereare strong arguments in favor of viewing theABO as the economic liability, on the groundsthat increases in the liability due to future salaryincreases accrue only when the increases areawarded (see, for example, Exley, Mehta, andSmith, 1997). Valuing and hedging of the ABOwith bonds thus forms the basis of a practicalrisk management approach.

Even if we consider projected benefit obliga-tions (PBOs), in many plans the proportion ofliabilities linked to future salary increases (andthe duration of the linkage, which does not usu-ally extend beyond retirement) may in practicebe quite small. Furthermore, to the extent thatliabilities are regarded as linked to future wages,the empirical evidence for a link between equi-ties and salary growth is weak (see Smith, 1998),as is the economic justification. (Even if a linkexisted between aggregate corporate earningsand wages, this does not necessarily imply a linkbetween earnings per share and wages peremployee.) More practically, no proposed linkshave met the acid test of a workable hedgingalgorithm. Although less accurate than the ABOhedge, inflation-linked bonds also provide thebest hedge for the salary-related element of aPBO liability.

Box 3.6. Defined Benefit (DB) Pensions and Corporate Finance Theory

Note: This box was prepared by Jon Exley ofMercer Investment Consulting.


112

The issue of asset allocation and pensionfunds has broad economic significance andraises important questions of long-run economicand financial relationships. In particular, itraises the question of whether a defined benefit(DB) pension scheme that is open and operatedon an ongoing basis ought to be 100 percentinvested in bonds, as some practitioners haveproposed, or whether equities should play a sub-stantial role, as is the case in most pension fundportfolios.

Salaries, profits, and dividends are all relativelystable components of GDP. As such, the estab-lished view has been that equities ought to be agood hedge against salary-linked liabilities (e.g.,Black, 1989, in the United States and Blake,2001, in the United Kingdom). Advocates ofbond investments question this. Some have evensuggested that the negative correlation of thetwo series shown in the first Figure means thatclaims on profits, such as equities, were not agood match for salaries (Exley, Mehta, andSmith, 1997). However, if the data are displayedin nominal terms over time (as in the secondFigure), the relationship becomes clearer. Tounderstand whether investment in equitiesshould form a standard part of a pension fund’sportfolio, the correct approach is to considerwhether changes in the value of future salariesare correlated with changes in the value offuture profits. Using rolling 25-year windows,Giles (2004) demonstrated that the correlationbetween changes in the present values of futureprofits and salaries was in excess of 80 percent.That is, holding a profit-linked security, like equi-ties, would have been a highly effective hedge forsalary-linked liabilities over the period.

Further arguments put forward in support ofthe proposition that funds should be 100 per-cent invested in bonds are also difficult tosustain:• Pension funds should hedge expected pro-

jected benefit obligations (PBOs), and notmerely the contractually certain accumulated

benefit obligations (ABOs). In general, firmshedge because future prices are not contractu-ally certain. Companies and pension fundsthat are operated as going concerns are con-cerned with expected liabilities, not just thosethat are contractually certain. This has funda-mental significance when it comes to assetallocation.

Box 3.7. Economics and Pension Fund Asset Allocation

Note: This box was prepared by Tim Giles ofCharles River Associates.

0

10

20

30

40

50

60

70

1948 53 58 63 68 73 78 83 88 93 98 2002

Salaries

Profits

United Kingdom: Salaries and Profits(In percent of GDP)

Source: United Kingdom, Office of National Statistics.

010

20

30

40

50

60

70

80

90

100

1948 53 58 63 68 73 78 83 88 93 98 2002

Nominal GDP

Salaryshare

Profits share

United Kingdom: Indexed NominalGDP Shares(1948 = 1)

Source: United Kingdom, Office of National Statistics.

Consultants have a significant influenceover pension fund management. Many pen-sion funds rely upon and commission theirrisk analysis to be done by external consult-ants. In addition, the corporate governancestructure of pension funds, with the overalldirection set by trustees or a benefits commit-tee, which may have limited expertise oninvestment matters, leads to considerablereliance on consultants for expert advice.Nevertheless, consultants often seem reluctantto propose substantial changes in ALM strate-gies or portfolio composition, particularly if itwould strongly deviate from their previousadvice or with consensus industry practices.Therefore, historically pension fund strategieshave been quite stable, but the reliance onconsensus presents some risk of herdingwithin the industry, and may retard the devel-opment of newer risk management practices.

Financial and risk management practiceswithin pension funds still focus much more

on the asset than the liability side of the bal-ance sheet. This is a distinct difference frominsurance companies, which traditionallyplaced much greater focus on liability riskrather than asset risk, as described in the April2004 GFSR. With pension funds, this partlyreflects the fact that assets are more easilyadjustable (particularly in the short term)than pension liabilities. The stronger empha-sis on assets also reflects the greater difficultyin recalculating liabilities than assets, with fullactuarial recalculations typically only per-formed once every three years and partialupdates (reviewing assumptions such as infla-tion, discount rates, and prospective invest-ment returns) only once a year. As such, thereis a tendency for funds to regard the value ofliabilities as fixed between actuarial revalua-tions, and this has created much less focus onliability risk.

Therefore, many pension funds measureasset performance against broad market


113

• Unless salaries are completely diversifiable ordeterministic, an economic valuation wouldnot rely on a bond rate. An economicallyimportant variable, such as salary levels, willbe highly correlated with systematic risk. Theappropriate discount rate is likely to be closeto the firm’s overall discount rate or, if that isunknown, the weighted average return on allasset classes rather than a bond rate.

• It is futile to promote the superiority of anasset allocation policy if it is based on securi-ties that are not available in sufficient quan-tity. When considering the matching of futuresalary obligations, 100 percent bond propo-nents suggest that pension funds should investin index-linked bonds—in spite of the lack ofsupply. They simply make the assumption thatgovernments or even corporations will allevi-ate the scarcity of index-linked bonds by issu-ing more. However, there is no obviousfundamental incentive for either to do so.

Accordingly, it is possible that the market forindex-linked liabilities could expand in theface of increased demand from pension fundsbut, in all likelihood, this would be at an equi-librium price that reflects an excess ofdemand over supply.The Capital Asset Pricing Model (CAPM) and

its extensions tell us that the notional “averageinvestor” should hold a portfolio equivalent inits risk/return characteristics to the “market” ofall assets. Pension funds do have peculiar char-acteristics, particularly DB schemes. Therefore,asset allocations may deviate from the marketportfolio. It has not yet been demonstrated,however, that the difference between the aver-age risk-reward trade-off, and that of pensionfund members or sponsors, is sufficiently starkto justify a 100 percent bond portfolio. In fact,the available evidence supports the widely heldview that equities are an important componentof the portfolios of open DB pension funds.

indices rather than against the underlying lia-bilities. But the focus on such indices may dis-tract managers’ attention from the real goal(i.e., to ensure funds are managed to meet lia-bilities as they fall due), and may furtherincrease the risk of herding. Moreover, insome jurisdictions, this passive investmentstyle also led pension fund managers to pur-sue very short-term investment strategies,including chasing yesterday’s attractive mar-kets and returns.

Greater focus on ALM and risk manage-ment practices needs to be encouraged.There has been increased use of ALM bysome pension funds, driven partly by the“perfect storm” of recent years, but also bygreater sensitivity to market values and move-ments through new accounting standards andmarket-related discount rates for liabilities.Some of the more sophisticated pensionfunds have begun to use risk managementtechniques from other areas of finance,instead of a focus primarily on actuarialmethods. Even if liabilities are recalculatedinfrequently, these funds now may employstochastic approaches such as Monte Carlosimulations to study a variety of different sce-narios and their impact on both the asset andliability side of the pension balance sheet. Butthere is much still to be done to encourageand employ ALM techniques more broadly inthe pension industry.

Several types of risks must be addressed inthe pension ALM process. Pension fund liabil-ities can vary through structural changes inthe workforce, the longevity of pensioners,future salary increases for workers, and theindexation of payments to prices or wages.Structural changes in the workforce are hardto hedge through the asset portfolio.However, annuities can hedge longevity risk,and salary and inflation-protected pensionbenefits may be in large part hedged by index-linked bonds.

Index-linked securities may be very usefulfor risk management but, as noted, the supplyof such instruments and their role in pensionportfolios have been limited to date. Theirlimited use in portfolios partly reflects the lim-ited indexation of pension obligations insome countries, and funding methodologiesthat do not fully reflect changes in inflationexpectations. Of course, the current low realyields available on index-linked securities(especially since most of the supply is of gov-ernment debt) may also limit demand. Butchanges in accounting and regulatory princi-ples could increase demand, and this could inturn stimulate increased supply from a widerrange of issuers, including corporates.

A modest allocation to alternative invest-ments may also play a useful role in pensionfund portfolios. Currently, many pensionfunds have alternative investment allocationsof around 10 to 15 percent, primarily in pri-vate equity and real estate. Hedge funds arealso being increasingly considered, although(despite some media reports) aggregateamounts invested, or reasonably expected tobe invested in the immediate future, remainmodest, and few pension funds have madeallocations to hedge funds above 5 to 10 per-cent of their investment portfolio.35

Increased international investment bringschallenges as well as benefits for risk manage-ment. As described earlier, internationalinvestments may provide both diversificationand higher returns. However, they will alsorequire a greater focus on currency risks andcredit risks to a wider range of countries, bor-rowers, and instruments, as well as broaderprogress on addressing obstacles to interna-tional capital mobility.

Pension funds should be free to choosetheir desired asset mix, within the bounds ofprudent and (ideally) risk-based funding prin-ciples. A degree of overfunding can provide aprudent cushion against the risk of market


114

35See, for example, Astérias (2002).

movements. Merely aiming to achieve a fund-ing ratio of 100 percent makes a pensionfund’s future level of contributions or benefitsmore sensitive to investment and market risk.Pension funds can increase the likelihood thattheir assets will be able to meet their liabilitiesby aiming to achieve a prudent measure ofoverfunding—the amount of which can bethought of as the “capitalization” of thefund—taking into account the level of riskarising from the asset mix and the financialstrength of the sponsor company.

Within this analysis, the financial strength ofthe sponsor company is an important (andoften overlooked) factor when consideringALM or risk management of pension funds.The strength of the sponsor is important inassessing the ability of a pension fund to meetits PBOs as a going concern (while thestrength of the pension fund as a stand-aloneentity may be more important to assess theability to meet ABOs in the event of a plan clo-sure). Financially strong companies in growingindustries may have more flexibility to takeinvestment risk and manage short-term fund-ing shortfalls. On the other hand, older-indus-try firms with a higher proportion of retireesto active workers, and with large pension obli-gations in relation to the size of the overallcompany or with less dependable cash flows,will have less flexibility to increase contribu-tions as needed. Therefore, we believe theappropriate funding level, or the risk profile ofa pension fund’s portfolio, should not be con-sidered in isolation from the financial strengthand flexibility of the sponsor company.

In sum, policymakers should seek to ensurethat a pension fund’s obligations can be metby its funding and investment strategy, consis-tent with its risk management practices.During our study, we frequently observed thestronger or “wealthier” sponsor companiesmoving to remove risk from their pensionfund and more often seeking to match oroverfund projected liabilities, while weakerfirms continued to pursue riskier investmentstrategies in the hope of “growing out” of

underfunded positions. In this respect, theexisting regulatory incentives and structurefor pension funds are producing very differ-ent behavior when compared to life insurers,many of whom reallocated from equities tobonds in response to weakened solvency posi-tions. Although this may in part also reflectthe longer time horizon and liability structureof pension funds, it suggests that risk-basedapproaches to funding and related regulationsmay be useful to encourage greater risk man-agement practices by pension funds.

Conclusions and PolicyRecommendations

The growth of funded pensions and the growingemphasis on risk management should strengthen therole of pension funds as stable, long-term institu-tional investors. Overall, this developmentshould enhance global financial stability.

Pension funds and their weakened financialposition have received significant attention inrecent years. No doubt, this can be attributedin large part to the 2000–02 equity marketdecline and falling interest rates. However, thedeeper causes of this deterioration have beenbuilding for many years. Nevertheless, pen-sion funds have a very significant role to playin mature market societies, particularly asproviders of retirement income and asinvestors of long-term savings. At present, anumber of factors challenge the very exis-tence of traditional pension structures inmany advanced economies, and the followingdiscussion and recommendations areintended to highlight how pension funds maycontinue to be a home for long-term savings,and thus an important contributor to socialand financial stability goals. The recent partialrecovery in funding ratios (arising fromimproved market conditions, particularly ris-ing market interest rates) provides a windowof opportunity for policymakers to introducemeasures to encourage better risk manage-ment practices, and to reduce the risk ofanother cycle of over- and underfunding.

CONCLUSIONS AND POLICY RECOMMENDATIONS

115

Promoting Sufficient Retirement Savings

As a first priority, policymakers need to more effec-tively communicate the pension and savingsagenda. In virtually every jurisdiction, publicand private sector officials we met highlightedthe need to better communicate the pensionchallenges and policy priorities. While thisseems universally the case, it is perhaps partic-ularly true in countries where the bulk of pen-sion benefits have been traditionally providedby the state. Indeed, in part due to insuffi-cient communication, some recent pensionreforms aimed at the household sector (Pillar3) have received little support or enthusiasm.The long-term need for greater savings is notgoing to dissipate, and for current pensionreform efforts to succeed a broad-basedunderstanding and support by the generalpublic is necessary.

Policymakers should provide effective incentivesfor the development of long-term savings. This doesnot require legislation or regulation that pro-vides detailed product design, but rather thedevelopment of a tax and legal environmentthat is relatively simple, stable, and facilitatesretirement savings growth. The private sectoris best equipped to design and provide a widevariety of savings products, and the public sec-tor should focus on building the necessaryframework and incentives. If the incentivesare properly established and communicated,we believe attractive products will emerge.

In designing a multi-pillar approach to pensionprovision, policymakers may be best served by tar-geting a relatively balanced contribution from eachpillar. With demographic and cost pressuresincreasing on Pillar 1, the contribution ofstate plans is projected much lower in mostadvanced economies. Increasingly, many statepension programs see their goal as providinga much lower or even minimum level ofreplacement income. Therefore, efforts tofacilitate larger contributions from Pillars 2and 3 are a practical necessity. As such, therole of retirement savings through occupa-tional pension schemes (Pillar 2) and/orindividual savings schemes (Pillar 3) will need

to grow significantly, as individuals seek tosupplement state benefits. However, due pri-marily to differing national preferences forrisk sharing between sectors, these pensionand savings programs will likely be designedvery differently.

As part of pension reform efforts, the workplace(Pillar 2) would seem to be the most efficient loca-tion to organize and accumulate retirement savings.Through occupational pension schemes,employers can most effectively organize thefunding of employees’ retirement savings.Moreover, employees seem more prepared tocontribute wages at source to long-term, work-related, pension schemes, whereas efforts toattract funds in various Pillar 3 schemes inmany mature market countries have experi-enced less success. In addition, by bundlingemployee savings and creating a menu offinancial products, employers are well posi-tioned to negotiate lower investment costs andobtain professional advice to the benefit ofemployees and beneficiaries.

Traditional DB schemes and principles shouldnot be uniformly discarded, and we believe thedevelopment of hybrid plans should be encouraged.Rather than being a flawed concept, many tra-ditional DB plans and benefits were mispricedand lacked adequate funding strategies andrisk management practices, as revealed by therecent market slump. Nevertheless, at leastsome of the risks related to pensions may bebetter managed at the institutional than theindividual level, and various hybrid plans thataim to guarantee a minimum level of benefitand corporate pension contribution, whilesharing some (not all) of the investment andlongevity risks with employees, may strike theright balance. A particular concern of manysponsors and industry analysts is longevity risk,and special consideration for the extreme eld-erly may support the broader market availabil-ity of annuities and related insuranceproducts. At the same time, Pillar 2 schemesmust be suitable for a more mobile workforce,including portability, proportionate benefitsand vesting schedules.


116

Promoting Strengthened RiskManagement Practices

Policymakers should consider ways to facilitatethe development of certain markets, including morelong-term (20 years and longer) fixed-income andindex-linked products. Such securities and mar-kets are necessary to allow pension funds tobetter match assets and liabilities, as well as tofacilitate the supply and pricing of annuityand long-term savings products by traditionalmarket participants, such as insurance compa-nies. We believe public sector leadership inthis area (including issuance) will be followedby greater private sector issuance. In severaljurisdictions, the number of institutions pro-viding annuity products continues to decline,and pricing is increasingly unattractive orunavailable due to the limited supply of mar-ket instruments to hedge such risk.

Financial stability can be enhanced by regulatorypolicies that are more closely aligned with the pur-pose and liability structure of pension funds, whileencouraging the development of better risk manage-ment systems. Regulators should encouragefunded plans to develop investment portfolios(including international investments) appro-priate to the pension’s liability structure. Suchmeasures would encourage fund managers tofocus more on risk management, rather thanbenchmarking performance against variousindices, and should also reduce the risk ofherd behavior. This may imply quite differentallocations between equities, bonds, and otherassets by different pension funds and, possibly,more fixed-income investments by pensionfunds with a rapidly aging workforce or closedto new participants.

Tax and related regulations should be designed toreduce or remove barriers to prudent, continuousfunding policies. One of the key reasons for thecontribution holidays in the late 1990s was theloss of preferential tax treatment (i.e., deduc-tions) or even tax penalties applied to furthercontributions once pension funds becamesomewhat overfunded. The inability to contin-uously fund and to build a funding cushionleft many funds exposed to a market down-

turn, and created the need for relatively largecontributions to meet minimum funding stan-dards. Tax rules that would allow a certainlevel of annual contributions, including as taxdeductible payments, even during overfundedperiods, and which do not penalize firms forbuilding up a prudent funding cushion (e.g.,two or three years of normal contributions),would help to encourage long-term, stablepension strategies. Moreover, based on OECDstatistics, such policies should not represent amaterial drain on tax revenue. Of course, abalance has to be reached to prevent pensionfunds from becoming tax shelters, and thresh-olds for continued tax deductibility could per-haps be coordinated with risk-based conceptsof adequate funding levels set by supervisors.

Risk-based approaches to supervision and toguarantee fund premiums should be enhanced.Guarantee fund “risk-based” premiums needto take account of the pension fund asset-lia-bility mix, and not only the level of currentfunding. This should provide a fairer distribu-tion of guarantee funds’ cost, reduce moralhazard risk, and encourage better risk man-agement practices. The Dutch proposal forrisk-based standards is an interesting andinnovative approach, applying supervisoryexpertise from other financial sectors. Itwould seem useful to also consider the finan-cial strength of the sponsor in setting risk-based capital or premiums; however, werecognize the practical difficulties, and weencourage the adoption of risk-basedapproaches whether or not that factor isincluded.

Policymakers and standard setters should ensurethat financial accounts provide an accurate reflec-tion of the financial condition of companies,including their pension plans, and we continue toencourage enhanced disclosure standards ratherthan an emphasis on single-point accounting meas-ures. Our recommended approach here is sim-ilar to that expressed in the April 2004 GFSRfor insurance companies. A factor frequentlycited by pension fund managers for the moveaway from DB plans is the trend to fair value

CONCLUSIONS AND POLICY RECOMMENDATIONS

117

accounting principles in many mature marketjurisdictions. While this view may understatethe demographic and cost pressures also atwork against DB plans, it is not clear that thevolatility associated with fair value accountingmeasures accurately reflects a pension fund’strue risk profile or properly focuses the man-agement of pension risks. We believe abroader disclosure of the asset and liabilitystructure (including the maturity profile ofpension obligations, and market and interestrate sensitivities) of funded pension plans,and a discussion of risk management practicesand funding or capital cushions, would pro-vide investors and beneficiaries with appropri-ate information. Indeed, while we support theapproach of rating agencies to treat theunfunded portion of pension obligations likeother forms of corporate debt, the agenciesacknowledge that such accounting volatilitycreates its own ratings pressure and possiblymore immediate funding constraints. Sophisti-cated investors are certainly aware of pensionissues, and relevant disclosure should ensurebroader market understanding of pensionrisks.

An important contribution to pension reform andthe growth of long-term savings may include the pro-motion of international diversification of pensionassets. In time, a shift of capital from advancedeconomies to younger and faster-growingeconomies may provide substantial benefits interms of higher returns and diversificationand, ultimately, in helping advancedeconomies deal with the macroeconomicimplications of aging. This reinforces theneed of policymakers to address the numer-ous frictions that continue to limit interna-tional capital mobility, and the need tostrengthen the capacity of developing coun-tries to absorb such potential capital flows.

From a financial stability perspective, pensionfunds represent a truly long-term institutionalinvestor base. However, following the 2000–02market downturn and low interest rate envi-ronment, which many analysts have called a“perfect storm” for pensions, we observe a sig-

nificant effort by sponsors to lower the riskprofile of their pension funds and to shift avariety of risks to pension beneficiaries (i.e.,the household sector). It seems increasinglyclear that households and individuals can beexpected to have a greater responsibility forsecuring their retirement, deciding how muchto save, where and how to invest, and toincreasingly bear other risks related to theirpensions and retirement. This risk transferraises the question of how well equippedhouseholds are to bear such risks, as well asthe appropriate sharing of risks between thehousehold and other sectors. These questionswill be addressed further in the March 2005Global Financial Stability Report chapter on thefund management industry and the house-hold sector in general.

ReferencesAllianz Dresdner Asset Management, 2003,

“European Pensions: Reform Trends and GrowthOpportunities” (Munich).

Aon Consulting, 2004, “Pension Crisis OverstatedSays Leading Pensions Consultancy,” pressrelease (London, June 22).

Astérias Ltd., 2002, U.K. Pension Fund Survey—August 2002 (London, October).

Black, Fischer, 1989, “Should You Use Stocks toHedge Your Pension Liability?” FinancialAnalysts Journal, Vol. 22 (January/February),pp. 10–12.

Blake, David, 2001, “U.K. Pension FundManagement: How Is Asset AllocationInfluenced by the Valuation of Liabilities?”Discussion Paper PI-0104 (London: The PensionInstitute, University of London).

Börsch-Supan, Axel, 2004, “Mind the Gap: TheEffectiveness of Incentives to Boost RetirementSaving in Europe,” Discussion Paper No. 52–04(Mannheim Research Institute for theEconomics of Aging, June).

Committee on Investment of Employee BenefitAssets (CIEBA), 2004, “The U.S. PensionCrisis—Evaluation and Analysis of EmergingDefined Benefit Pension Issues” (Bethesda,MD: Association for Financial Professionals,March).


118

Custis, Thomas, 2001, “Defined Benefit PensionPlans Face the ‘Perfect Storm’” (Seattle:Milliman USA, December 19).

Deutsche Bank Research, 2003, “Aging, theGerman Rate of Return and Global CapitalMarkets,” Current Issues: Demography Special(Frankfurt: December 4).

Exley, C. Jon, Shyam J. B. Mehta, and Andrew D.Smith, 1997, “The Financial Theory of DefinedBenefit Pension Schemes,” Group for Economicand Market Value Based Studies, pp. 72–81. Avail-able via the Internet at http://www.gemstudy.com/DefinedBenefitPensionsDownloads/Financial_Theory_of_Defined_Benefit_Pension_Schemes.pdf.

Federal Deposit Insurance Corporation, 2004,“Could a Bull Market Be a Panacea for DefinedBenefit Pension Plans?” (Washington: FDIC).

Galer, Russell, 2002, “Prudent Person RuleStandard for the Investment of Pension FundAssets” (Paris: Organisation for Economic Co-operation and Development, November).

Giles, Tim, 2004, “The 100% Bonds Proposition:Does the Underlying Analysis Stand Up?”(London: Charles River Associates, April).Available via the Internet at http://www.crai.co.uk/pubs/pub_3653.pdf.

Greenwich Associates, 2003, “As Funding GapsResist Rally, U.S. Investment Plans Shop forSmarter Strategies” (Greenwich, CT, February 9).

Hewitt Investment Group, 2001, “The PerfectStorm: How Pension Funds Should NavigateThrough Troubled Waters” (Lincolnshire, IL:October).

Higgins, Matthew, 1998, “Demography, NationalSavings, and International Capital Flows,”International Economic Review, Vol. 39 (May),pp. 343–69.

International Monetary Fund, 2003, “Underfundingof Corporate Pension Plans: Macroeconomic andPolicy Implications,” in United States: Selected Issues,Country Report No. 03/245 (Washington: IMF).Available via the Internet at http://www.imf.org/external/pubs/cat/longres.cfm?sk=16785.0.

———, 2004a, Global Financial Stability Report,World Economic and Financial Surveys(Washington: IMF, April).

———, 2004b, “Pension Reform Issues in Japan,”in Japan: Selected Issues, Country Report No.04/247 (Washington: IMF). Available via theInternet at http://www.imf.org/external/pubs/cat/longres.cfm?sk=17621.0.

Jackson, Patricia, William Perraudin, andKamakshya Trivedi, forthcoming, Defined BenefitPensions, Hedging and UK Companies (London:Bank of England).

Lührmann, Melanie, 2002, “The Role ofDemographic Change in ExplainingInternational Capital Flows,” MEA-DiscussionPaper (Universität Mannheim).

Modigliani, Franco, and Merton H. Miller, 1958,“The Cost of Capital, Corporation Finance, andthe Theory of Investment,” American EconomicReview, Vol. 48 (June), pp. 261–97.

Moody’s Investors Service, 2003, “AnalyticalObservations Related to U.S. PensionObligations” (January).

———, 2004, “The Effects of Aging on PublicSector Pensions and Healthcare Systems: ARating Agency Perspective” (May).

Organisation for Economic Co-operation andDevelopment, 2004a, “Recommendation onCore Principles of Occupational PensionRegulation” (Paris: OECD)

——— 2004b, “Tax Favoured Retirement SavingsPlans: A Review of Budgetary Implications andPolicy Issues,” Working Paper No. 1 on Macro-economic and Structural Policy Analysis,Economics Department, Economic PolicyCommittee (Paris: OECD).

Queisser, Monika, and Dimitri Vittas, 2000,“The Swiss Multi-Pillar Pension System:Triumph of Common Sense?” Policy ResearchWorking Paper No. 2416 (Washington: WorldBank).

Reagan, Patricia, and John Turner, 2000, “Didthe Decline in Marginal Tax Rates During the1980s Reduce Pension Coverage?” in EmployeeBenefits and Labor Markets in Canada and theUnited States, ed. by William T. Alpert andStephen A. Woodbury (Kalamazoo, MI: W.E.Upjohn Institute).

Reisen, Helmut, 2000, Pensions, Savings and CapitalFlows: From Ageing to Emerging Markets(Cheltenham, U.K.: Edward Elgar).

Schieber, Sylvester J., 2003, “Global Aging: Oppor-tunity or Threat for the U.S. Economy?” testi-mony before the U.S. Senate Special Committeeon Aging (Washington, February 27).

Shilling, A. Gary, 2003, “Pension Profits BecomeCorporate Costs,” Business Economics (October).

Smith, Andrew D., 1998, “Salary Related CashFlows: Market Based Valuation,” Group for Eco-

REFERENCES

119

nomic and Market Value Based Studies. Availablevia the Internet at http://www.gemstudy.com/DefinedBenefitPensionsDownloads/Salary_Related_Cash_Flows_Market_Based_Valuation.pdf.

Standard & Poor’s, 2003, “Navigating the Inter-national Pension Accounting Maze” (May 16).

———, 2004a, “Update on Postretirement BenefitsObligations” (April 29).

———, 2004b, “Using Equity Duration in PensionFund Allocation” (January 27).

Turner, Adair, 2003, “Demographics, Economicsand Social Choice,” transcript of lecture atLondon School of Economics (November 6).

Van Ewijk, Casper, and Martijn van de Ven, 2003,“Pension Funds at Risk,” CBP Report 2003/1(The Hague: Central Plaanbureau, April).

Watson Wyatt, 2003, “Pensions in Crisis,” Insider,Vol. 13, No. 9 (September).

World Bank, 1994, Averting the Old Age Crisis: Policiesto Protect the Old and Promote Growth, World BankPolicy Research Reports (New York: OxfordUniversity Press).

Yermo, Juan, 2003, “Survey of InvestmentRegulation of Pension Funds” (Paris:Organisation for Economic Co-operation andDevelopment, November).


120

risk management and the pension fund industry · such strategies are not only driven by funda- ......

Documents